black_jack_neural_net_v4.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import random\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import sklearn.metrics as metrics\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense, LSTM, Flatten, Dropout"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# This function lists out all permutations of ace values in the array sum_array\n",
    "# For example, if you have 2 aces, there are 4 permutations:\n",
    "#     [[1,1], [1,11], [11,1], [11,11]]\n",
    "# These permutations lead to 3 unique sums: [2, 12, 22]\n",
    "# Of these 3, only 2 are <=21 so they are returned: [2, 12]\n",
    "def get_ace_values(temp_list):\n",
    "    sum_array = np.zeros((2**len(temp_list), len(temp_list)))\n",
    "    # This loop gets the permutations\n",
    "    for i in range(len(temp_list)):\n",
    "        n = len(temp_list) - i\n",
    "        half_len = int(2**n * 0.5)\n",
    "        for rep in range(int(sum_array.shape[0]/half_len/2)):\n",
    "            sum_array[rep*2**n : rep*2**n+half_len, i] = 1\n",
    "            sum_array[rep*2**n+half_len : rep*2**n+half_len*2, i] = 11\n",
    "    # Only return values that are valid (<=21)\n",
    "    # return list(set([int(s) for s in np.sum(sum_array, axis=1) if s<=21]))\n",
    "    return [int(s) for s in np.sum(sum_array, axis=1)]\n",
    "\n",
    "# Convert num_aces, an int to a list of lists\n",
    "# For example if num_aces=2, the output should be [[1,11],[1,11]]\n",
    "# I require this format for the get_ace_values function\n",
    "def ace_values(num_aces):\n",
    "    temp_list = []\n",
    "    for i in range(num_aces):\n",
    "        temp_list.append([1,11])\n",
    "    return get_ace_values(temp_list)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Make a deck\n",
    "def make_decks(num_decks, card_types):\n",
    "    new_deck = []\n",
    "    for i in range(num_decks):\n",
    "        for j in range(4):\n",
    "            new_deck.extend(card_types)\n",
    "    random.shuffle(new_deck)\n",
    "    return new_deck\n",
    "\n",
    "# Total up value of hand\n",
    "def total_up(hand):\n",
    "    aces = 0\n",
    "    total = 0\n",
    "    \n",
    "    for card in hand:\n",
    "        if card != 'A':\n",
    "            total += card\n",
    "        else:\n",
    "            aces += 1\n",
    "    \n",
    "    # Call function ace_values to produce list of possible values for aces in hand\n",
    "    ace_value_list = ace_values(aces)\n",
    "    final_totals = [i+total for i in ace_value_list if i+total<=21]\n",
    "    \n",
    "    if final_totals == []:\n",
    "        return min(ace_value_list) + total\n",
    "    else:\n",
    "        return max(final_totals)\n",
    "    \n",
    "# Play a game of blackjack (after the cards are dealt)\n",
    "def play_game(dealer_hand, player_hands, blackjack, curr_player_results, dealer_cards, hit_stay, card_count, dealer_bust):\n",
    "    action = 0\n",
    "    # Dealer checks for 21\n",
    "    if set(dealer_hand) == blackjack:\n",
    "        for player in range(players):\n",
    "            if set(player_hands[player]) != blackjack:\n",
    "                curr_player_results[0,player] = -1\n",
    "            else:\n",
    "                curr_player_results[0,player] = 0\n",
    "    else:\n",
    "        for player in range(players):\n",
    "            # Players check for 21\n",
    "            if set(player_hands[player]) == blackjack:\n",
    "                curr_player_results[0,player] = 1\n",
    "            else:\n",
    "                # Hit randomly, check for busts\n",
    "                if (hit_stay >= 0.5) and (total_up(player_hands[player]) != 21):\n",
    "                    player_hands[player].append(dealer_cards.pop(0))\n",
    "                    card_count[player_hands[player][-1]] += 1\n",
    "                    \n",
    "                    action = 1\n",
    "                    live_total.append(total_up(player_hands[player]))\n",
    "                    if total_up(player_hands[player]) > 21:\n",
    "                        curr_player_results[0,player] = -1\n",
    "\n",
    "    # Dealer hits based on the rules\n",
    "    card_count[dealer_hand[-1]] += 1\n",
    "    while total_up(dealer_hand) < 17:\n",
    "        dealer_hand.append(dealer_cards.pop(0))\n",
    "        card_count[dealer_hand[-1]] += 1\n",
    "        \n",
    "    # Compare dealer hand to players hand but first check if dealer busted\n",
    "    if total_up(dealer_hand) > 21:\n",
    "        dealer_bust.append(1)\n",
    "        for player in range(players):\n",
    "            if curr_player_results[0,player] != -1:\n",
    "                curr_player_results[0,player] = 1\n",
    "    else:\n",
    "        dealer_bust.append(0)\n",
    "        for player in range(players):\n",
    "            if total_up(player_hands[player]) > total_up(dealer_hand):\n",
    "                if total_up(player_hands[player]) <= 21:\n",
    "                    curr_player_results[0,player] = 1\n",
    "            elif total_up(player_hands[player]) == total_up(dealer_hand):\n",
    "                curr_player_results[0,player] = 0\n",
    "            else:\n",
    "                curr_player_results[0,player] = -1\n",
    "                \n",
    "    return curr_player_results, dealer_cards, action, card_count, dealer_bust"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 194,
   "metadata": {},
   "outputs": [],
   "source": [
    "stacks = 40000\n",
    "players = 1\n",
    "num_decks = 1\n",
    "\n",
    "card_types = ['A',2,3,4,5,6,7,8,9,10,10,10,10]\n",
    "\n",
    "dealer_card_feature = []\n",
    "player_card_feature = []\n",
    "player_live_total = []\n",
    "player_live_action = []\n",
    "player_results = []\n",
    "dealer_bust = []\n",
    "\n",
    "first_game = True\n",
    "prev_stack = 0\n",
    "stack_num_list = []\n",
    "new_stack = []\n",
    "card_count_list = []\n",
    "games_played_with_stack = []\n",
    "\n",
    "for stack in range(stacks):\n",
    "    games_played = 0\n",
    "    \n",
    "    # Make a dict for keeping track of the count for a stack\n",
    "    card_count = {2: 0,\n",
    "                  3: 0,\n",
    "                  4: 0,\n",
    "                  5: 0,\n",
    "                  6: 0,\n",
    "                  7: 0,\n",
    "                  8: 0,\n",
    "                  9: 0,\n",
    "                  10: 0,\n",
    "                  'A': 0}\n",
    "    \n",
    "    blackjack = set(['A',10])\n",
    "    dealer_cards = make_decks(num_decks, card_types)\n",
    "    while len(dealer_cards) > 20:\n",
    "        \n",
    "        curr_player_results = np.zeros((1,players))\n",
    "        \n",
    "        dealer_hand = []\n",
    "        player_hands = [[] for player in range(players)]\n",
    "        live_total = []\n",
    "        live_action = []\n",
    "\n",
    "        # Deal FIRST card\n",
    "        for player, hand in enumerate(player_hands):\n",
    "            player_hands[player].append(dealer_cards.pop(0))\n",
    "            card_count[player_hands[player][-1]] += 1\n",
    "            \n",
    "        dealer_hand.append(dealer_cards.pop(0))\n",
    "        card_count[dealer_hand[-1]] += 1\n",
    "        \n",
    "        # Deal SECOND card\n",
    "        for player, hand in enumerate(player_hands):\n",
    "            player_hands[player].append(dealer_cards.pop(0))\n",
    "            card_count[player_hands[player][-1]] += 1\n",
    "            \n",
    "        dealer_hand.append(dealer_cards.pop(0))\n",
    "        \n",
    "        # Record the player's live total after cards are dealt\n",
    "        live_total.append(total_up(player_hands[player]))\n",
    "        \n",
    "        if stack < stacks/2:\n",
    "            hit_stay = 1\n",
    "        else:\n",
    "            hit_stay = 0\n",
    "            \n",
    "        curr_player_results, dealer_cards, action, card_count, dealer_bust = play_game(dealer_hand, player_hands, \n",
    "                                                                          blackjack, curr_player_results, \n",
    "                                                                          dealer_cards, hit_stay, card_count,\n",
    "                                                                          dealer_bust)\n",
    "        \n",
    "        # Track features\n",
    "        dealer_card_feature.append(dealer_hand[0])\n",
    "        player_card_feature.append(player_hands)\n",
    "        player_results.append(list(curr_player_results[0]))\n",
    "        player_live_total.append(live_total)\n",
    "        player_live_action.append(action)\n",
    "        \n",
    "        # Update card count list with most recent game's card count\n",
    "        if stack != prev_stack:\n",
    "            new_stack.append(1)\n",
    "        else:\n",
    "            new_stack.append(0)\n",
    "            if first_game == True:\n",
    "                first_game = False\n",
    "            else:\n",
    "                games_played += 1\n",
    "            \n",
    "        stack_num_list.append(stack)\n",
    "        games_played_with_stack.append(games_played)\n",
    "        card_count_list.append(card_count.copy())\n",
    "        prev_stack = stack\n",
    "\n",
    "model_df = pd.DataFrame()\n",
    "model_df['dealer_card'] = dealer_card_feature\n",
    "model_df['player_total_initial'] = [total_up(i[0][0:2]) for i in player_card_feature]\n",
    "model_df['hit?'] = player_live_action\n",
    "\n",
    "has_ace = []\n",
    "for i in player_card_feature:\n",
    "    if ('A' in i[0][0:2]):\n",
    "        has_ace.append(1)\n",
    "    else:\n",
    "        has_ace.append(0)\n",
    "model_df['has_ace'] = has_ace\n",
    "\n",
    "dealer_card_num = []\n",
    "for i in model_df['dealer_card']:\n",
    "    if i=='A':\n",
    "        dealer_card_num.append(11)\n",
    "    else:\n",
    "        dealer_card_num.append(i)\n",
    "model_df['dealer_card_num'] = dealer_card_num\n",
    "\n",
    "model_df['Y'] = [i[0] for i in player_results]\n",
    "lose = []\n",
    "for i in model_df['Y']:\n",
    "    if i == -1:\n",
    "        lose.append(1)\n",
    "    else:\n",
    "        lose.append(0)\n",
    "model_df['lose'] = lose\n",
    "\n",
    "correct = []\n",
    "for i, val in enumerate(model_df['lose']):\n",
    "    if val == 1:\n",
    "        if player_live_action[i] == 1:\n",
    "            correct.append(0)\n",
    "        else:\n",
    "            correct.append(1)\n",
    "    else:\n",
    "        if player_live_action[i] == 1:\n",
    "            correct.append(1)\n",
    "        else:\n",
    "            correct.append(0)\n",
    "model_df['correct_action'] = correct"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 195,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Make a new version of model_df that has card counts\n",
    "card_count_df = pd.concat([pd.DataFrame(new_stack, columns=['new_stack']),\n",
    "                           pd.DataFrame(games_played_with_stack, columns=['games_played_with_stack']),\n",
    "                           pd.DataFrame.from_dict(card_count_list),\n",
    "                           pd.DataFrame(dealer_bust, columns=['dealer_bust'])], axis=1)\n",
    "model_df = pd.concat([model_df, card_count_df], axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 196,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAsgAAAGoCAYAAABbtxOxAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvOIA7rQAAIABJREFUeJzt3XmYJWV99vHvzaCCwqsi4xIWB5RLJcR1QIgK7gEXcA8QcUPRRNyjUUwQifq6K74SI6IRFUTADZWAG4tLBAZUFEZkBJURhBFERCLr7/2jaqRsuk+fcfpUzUx/P9fVV59TVV3nPuU43FP9nOdJVSFJkiSpsd7QASRJkqQ1iQVZkiRJ6rAgS5IkSR0WZEmSJKnDgixJkiR1WJAlSZKkDguyJEmS1GFBliRJkjosyJIkSVLH+kMHmAubbrppLVq0aOgYkiRJWoOdddZZv6mqhbMdt04U5EWLFrFkyZKhY0iSJGkNluQX4xznEAtJkiSpw4IsSZIkdViQJUmSpA4LsiRJktRhQZYkSZI6LMiSJElShwVZkiRJ6rAgS5IkSR0WZEmSJKnDgixJkiR19F6Qk+ya5Pwky5K8fpr9z0uyIskP2q8X9p1RkiRJ89f6fb5YkgXAocDjgOXAmUmOr6rzphz6marav89skiRJEvR/B3kHYFlVXVhV1wNHA3v0nEGSJEmaUd8FeTPg4s7z5e22qZ6e5JwkxyXZYroTJdkvyZIkS1asWDGJrJIkSZqH+i7ImWZbTXn+JWBRVd0f+DpwxHQnqqrDqmpxVS1euHDhHMeUJEnSfNV3QV4OdO8Ibw5c0j2gqq6oquvapx8BHtJTNkmSJKnfD+kBZwLbJNkK+BWwJ7B394Ak96iqS9unuwNL+40oaV32wdd8aegIc2r/9zx56AiStM7ptSBX1Y1J9gdOAhYAH6uqc5McDCypquOBlyfZHbgRuBJ4Xp8ZJUmSNL/1fQeZqjoBOGHKtgM7j98AvKHvXNJ8cOrOuwwdYU7tctqpQ0eQJK2DXElPkiRJ6rAgS5IkSR0WZEmSJKnDgixJkiR1WJAlSZKkDguyJEmS1GFBliRJkjosyJIkSVKHBVmSJEnqsCBLkiRJHRZkSZIkqWP9oQNIfXjY/3vY0BHm1Hde9p2hI0iStM7yDrIkSZLUYUGWJEmSOizIkiRJUocFWZIkSeqwIEuSJEkdFmRJkiSpw4IsSZIkdViQJUmSpA4LsiRJktRhQZYkSZI6LMiSJElSx/pDB9Dk/fLgvxk6wpza8sAfDR1BkiStw7yDLEmSJHVYkCVJkqQOC7IkSZLUYUGWJEmSOizIkiRJUocFWZIkSeqwIEuSJEkdFmRJkiSpw4IsSZIkdazzK+k95LWfGDrCnDrrXc8ZOoIkSdI6zTvIkiRJUocFWZIkSeqwIEuSJEkdFmRJkiSpw4IsSZIkdViQJUmSpI51fpo3SdKfe+uznzF0hDn1xk8dN3QESesY7yBLkiRJHRZkSZIkqcOCLEmSJHVYkCVJkqQOC7IkSZLUYUGWJEmSOizIkiRJUocFWZIkSeqwIEuSJEkdFmRJkiSpw4IsSZIkdViQJUmSpA4LsiRJktRhQZYkSZI6LMiSJElShwVZkiRJ6rAgS5IkSR0WZEmSJKnDgixJkiR1WJAlSZKkDguyJEmS1LF+3y+YZFfgEGABcHhVvX2G454BHAtsX1VLeowoSVrHLX3rN4eOMOfu98ZHDx1BWmf0egc5yQLgUGA3YFtgryTbTnPcxsDLgdP7zCdJkiT1PcRiB2BZVV1YVdcDRwN7THPcvwPvBP7YZzhJkiSp74K8GXBx5/nydtufJHkQsEVVfbnPYJIkSRL0X5Azzbb6085kPeB9wGtmPVGyX5IlSZasWLFiDiNKkiRpPuu7IC8Htug83xy4pPN8Y2A74JQkPwd2BI5PsnjqiarqsKpaXFWLFy5cOMHIkiRJmk/6LshnAtsk2SrJbYE9geNX7qyq31XVplW1qKoWAd8DdncWC0mSJPWl14JcVTcC+wMnAUuBY6rq3CQHJ9m9zyySJEnSdHqfB7mqTgBOmLLtwBmOfWQfmSRJkqSVxi7ISbYGngVsCWwwZXdV1b5zGUySJEkawlgFOckeNKvarQdcDlw35ZC61Q9JkiRJa6Fx7yC/BTgF+Ieqck41SZIkrbPGLchbA6+xHEuSJGldN+4sFj8B7jLJIJIkSdKaYNyC/DrggPaDepIkSdI6a9whFgfR3EFemuQC4Mop+6uqdpnLYJIkSdIQxi3INwHnTzKIJEmStCYYqyC7YIckSZLmi16XmpYkSZLWdDPeQU6yM3B2VV3TPh6pqk6b02SSJGmiDjrooKEjzKl17f1oOKOGWJwC7Aic0T6eabW8tPsWzGUwSZIkaQijCvKjgPM6jyVJkqR13owFuapOne6xJEmStC6b8UN6Sd6Z5IlJ7thnIEmSJGlIo2axeBVwPHBFku8nOSTJ05Js2lM2SZIkqXejxiBvAjwM2AV4BPBi4GVAJTkfOBU4DTitqn416aCSJElSH0aNQf49cGL7RZINgJ2AnduvfYD92n0XVdW9J55WkiRJmrBxl5qmqv4InAycnGQBzd3lVwNPBraaTDxJkiSpX2MV5Pbu8Y7ccvd4R2ADYCnwYeBbkwooSZIk9WnUSnq7cksh3p5mMZCzgG8D7we+U1W/7SOkJEmS1JdRd5BPAP4AHA4cAJzRDrOQJEmS1lmjCvJJNB/KewXwOOBbSU4Dvl1VF/cRTpIkSerbqFksdkuyHvBAbhlqcQhwlyQX04w7/hbwrapa2kdYSZIkadJGfkivqm4Gzm6/3g+Q5K9p5kXeGfhXYLMkV1TVXSecVZIkSZq4USvpzeSSztdl7ba7zFkiSZIkaUCzTvOW5O7cMsRiZ2BbIMBNNLNavJtmVT1JkiRprTdqmrfDaQrxvWgK8XXAGcDbaJaY/m5VXdtHSEmSJKkvo+4g/z3wP8AnaO4Qn15V1/eSSpIkSRrIqIJ8p6q6qbckkiRJ0hpgxg/pWY4lSZI0H/0ls1hIkiRJ6ywLsiRJktRhQZYkSZI6Zi3ISRYkeUCShX0EkiRJkoY0zh3kApYAD5pwFkmSJGlws66kV1U3J7kYuEMPeSRJknpzzLE7DB1hTj3rmWcMHWGdMO4Y5A8Dr0xy20mGkSRJkoY26x3k1sY0S05fmORE4FKaoRcrVVW9aa7DSZIkSX0btyAf0Hn8gmn2F2BBliRJ0lpvrIJcVU4HJ0mSpHnB4itJkiR1jDvEAoAkTwJ2ATYBrgBOraqvTCKYJEmSNISxCnKSjYEvA48AbqQpx3cBXpPkW8CTquqaiaWUJEmSejLuEIu3AQ8G9gE2rKp7ABsCz2m3v20y8SRJkqR+jVuQnw78a1UdWVU3AVTVTVV1JPBv7X5JkiRprTduQb4LcN4M+85r90uSJElrvXEL8kXAk2bY94R2vyRJkrTWG3cWiw8D70myEXAkzUp6dwf2BF4IvHoy8SRJkqR+jbtQyPuSLAReBTyv3RzgOuDtVXXIZOJJkiRJ/Rp7HuSqOiDJu4AdaeZBvhL4XlX9dlLhJEmSpL6t0kIhbRn+7wllkSRJkgbnUtOSJElShwVZkiRJ6rAgS5IkSR0WZEmSJKnDgixJkiR1jF2Qk9whycuTHJfk5CTbtNv3THLfyUWUJEmS+jPWNG9JtgBOATYHfgJsB2zc7n4U8FiaFfUkSZKktdq4d5DfQ7Nq3jbAQ2hW0VvpVGDnOc4lSZIkDWLchUIeB+xXVb9MsmDKvl8Bm81tLEmSJGkY495Bvi3w+xn23RG4YW7iSJIkScMatyCfAzx9hn27AWeN+4JJdk1yfpJlSV4/zf6XJPlRkh8k+XaSbcc9tyRJkrS6xh1i8S7guCQAR7Xbtk2yB7AvsPs4J2mHZxxKM2RjOXBmkuOr6rzOYUdV1X+2x+8OvBfYdcyckiRJ0moZ6w5yVX0O+CfgmcDX282fAF4J7F9VJ475ejsAy6rqwqq6Hjga2GPKa13deXoHoMY8tyRJkrTaxr2DTFX9Z5JPAjsBdwWuAL5bVTONTZ7OZsDFnefLgYdOPSjJS4FX04x9fvQqnF+SJElaLWMXZICq+gO33EH+S2Sabbe6Q1xVhwKHJtkb+Ffgubc6UbIfsB/AlltuuRqRJEmSpFvMWJCT7AycXVXXtI9HqqrTxni95cAWneebA5eMOP5o4EMzvN5hwGEAixcvdhiGJEmS5sSoO8inADsCZ7SPZyqhafdNnR95OmcC2yTZimb+5D2Bvf/sZMk2VXVB+/SJwAVIkiRJPRlVkB8FnNd5vNqq6sYk+wMn0RTqj1XVuUkOBpZU1fHA/kkeSzO38m+ZZniFJEmSNCmjCvKf7hhX1alz9YJVdQJwwpRtB3Yev2KuXkuSJElaVaOmeTsZcJEOSZIkzSujCvJ0M05IkiRJ67Rxl5qWJEmS5oXZ5kFeL8m4q+3dPAd5JEmSpEHNVpC/M+Z5aoxzSZIkSWu82Urtx2gW95AkSZLmhdkK8keq6oxekkiSJElrAD+kJ0mSJHVYkCVJkqSOUQX5CGBFX0EkSZKkNcGMY5Cr6vl9BpEkSZLWBA6xkCRJkjosyJIkSVKHBVmSJEnqmLEgJ3lvki3axzsn2ai/WJIkSdIwRt1BfiVwj/bxycC2k48jSZIkDWtUQV4BbNM+DlCTjyNJkiQNa9RS018F/ivJW9vnX0hy3QzHVlXda26jSZIkSf0bVZBfDJxDM7TiOe3j3/QRSpIkSRrKqIVCrgXeBZDkucCbquqMvoJJkiRJQxh1B/lPqsrp4CRJkjQvjFWQAZLcHngBsAuwCXAFcArw8fZusyRJkrTWG+vOcJK7A2cDHwAWA7cHtgc+CJyV5G4TSyhJkiT1aNyhE+8E7gw8oqq2qqqdqmor4OHAnYB3TCqgJEmS1KdxC/JuwBuq6jvdjVX1XeBfgSfOdTBJkiRpCOMW5I2AS2bYt7zdL0mSJK31xi3I5wP7zLDv2cBP5iaOJEmSNKxxZ7F4N/CJ9sN4RwGXAncH9gQey8zlWZIkSVqrjDsP8qfaad4OBg7v7LoMeElVHTWJcJIkSVLfxp4HuaoOS3I4cB+aeZCvBM6vqpsnFU6SJEnq29gFGaAtw0snlEWSJEk9e8BxJw0dYc798Bl/t1o/7xLSkiRJUocFWZIkSeqwIEuSJEkdFmRJkiSpY6yCnOSOkw4iSZIkrQnGvYN8SZKPJtl+omkkSZKkgY1bkN8FPA74XpLvJ9kvyUYTzCVJkiQNYqyCXFUHAYuApwKXAP9Bc1f5Q0keOLF0kiRJUs/G/pBeVd1cVcdX1ROBewGHALsDZyU5PcnzktxuUkElSZKkPvyls1hcTbPU9DVAgDsCHwWWJXn4HGWTJEmSerdKBTnJw5J8AvgV8Gbgm8ADquq+wP2AC4EPz3lKSZIkqSfrj3NQkpcBL6YpwUuB1wKfqKrfrzymqn6a5E3ANyYRVJIkSerDWAUZeDfwBeClVXXqiOMuAA5e7VSSJEnSQMYtyFtW1WWzHVRVK4deSJIkSWulcccg/0+SB0y3I8l2SS6cw0ySJEnSYMYtyIuAmaZw2wC455ykkSRJkga2KrNY1AzbFwNXzUEWSZIkaXAzjkFO8irgVe3TAr6U5Poph20IbAIcPZl4kiRJUr9GfUjvQm6Zsu25wBJgxZRjrgPOAw6f+2iSJElS/2YsyFX1ReCLAEkADq6qi3rKJUmSJA1irGnequr5kw4iSZIkrQlGjUE+EDi8qi5pH49SVfXvcxtNkiRJ6t+oO8gHAScCl7SPRynAgixJkqS13qgxyOtN91iSJElal1l8JUmSpA4LsiRJktQx6kN6NzPz6nlTVVWNNSOGJEmStCYbVWoPZvyCLEmSJK0TRn1I76Aec0iSJElrBMcgS5IkSR0uFCJJkiR19L5QSJJdgUOABTQF/O1T9r8aeCFwI7ACeEFV/WKcc0uSJEmra8YhFlW1XlWd0Xk86mvBOC+WZAFwKLAbsC2wV5Jtpxz2fWBxVd0fOA5451/yxiRJkqS/RN9jkHcAllXVhVV1PXA0sEf3gKo6uaqubZ9+D9i854ySJEmax1Zp7uIkjwJ2AjYDfgX8T1WdvAqn2Ay4uPN8OfDQEcfvC/z3qmSUJEmSVsdYBTnJJsCxwCNpxhv/FrhzsyunAM+sqivHOdU026adaznJs4HFwC4z7N8P2A9gyy23HOOlJUmSpNmNO8TiA8D2wD7AhlW1ENgQeA5NiT1kzPMsB7boPN+c5kOAfybJY4E3ArtX1XXTnaiqDquqxVW1eOHChWO+vCRJkjTauEMsngy8oaqOWrmhqm4AjmzvLr9lzPOcCWyTZCuaIRp7Ant3D0jyIODDwK5VdfmY55UkSZLmxLh3kG8CLphh3/nt/llV1Y3A/sBJwFLgmKo6N8nBSXZvD3sXsBFwbJIfJDl+zIySJEnSahv3DvIXgb8HvjrNvj2BL4z7glV1AnDClG0Hdh4/dtxzSZIkSXNt1Ep6j+48/RLw/iRfofmw3mXA3YBnAX8NvGKSISVJkqS+jLqD/HWaGSbS+b45zSIfU32WZmU8SZIkaa02qiA/qrcUkiRJ0hpixoJcVaf2GUSSJElaE/S91LQkSZK0Rht7qekk29Es/XwfYIMpu6uqHjOXwSRJkqQhjLvU9EOBU4GfA9sA59AsNb0lzep4yyaUT5IkSerVuEMs3gZ8jmZKtwD7VtUi4LE0s1eMu5KeJEmStEYbtyDfH/gUzXRv0E7pVlXfpCnH/3fuo0mSJEn9G7cg3wb4Q1XdDFwJ3KOz73xgu7kOJkmSJA1h3IL8M2Cz9vE5wAuSrJdkPeD5wK8nEU6SJEnq27izWHwJeCRwFM145K8AVwM3ARsBL59EOEmSJKlvYxXkqjqo8/jrSXYCng5sCJxYVV+dTDxJkiSpX2PPg9xVVWcDZ89xFkmSJGlwq1SQk9wb2IFmPPKvgDOqyjmQJUmStM4Yd6GQDYD/APahneKtdVOSI4CXVtV1E8gnSZIk9WrcWSzeDfwD8Cbg3sDG7feDaErzuyYRTpIkSerbuEMs9gTeXFVv62y7EHhrEoBX4UwWkiRJWgeMewf5dsAZM+w7Hbjt3MSRJEmShjVuQf468PgZ9j0e+ObcxJEkSZKGNeMQiyRbd56+F/hkkjsAxwKXAXcDngU8AXj2JENKkiRJfRk1BnkZUJ3nAf4ReMmUbQCn8uezW0iSJElrpVEF+fm9pZAkSZLWEDMW5Ko6os8gkiRJ0ppgVVfSC7AtsAlwBbC0qmr0T0mSJElrj3FnsSDJC4FLgXOAU4AfAZck2Xcy0SRJkqT+jbvU9D8AhwHfAD4F/Bq4O83qeoclubaqPj2xlJIkSVJPxh1i8TrgyKraZ8r2I5J8EvgXwIIsSZKktd64QyzuQ3PneDqfavdLkiRJa71xC/Lvgc1n2Ld5u1+SJEla641bkP8beFuSR3Q3JtkJeEu7X5IkSVrrrcoY5B2BU5L8imY2i7vT3D1e1u6XJEmS1npjFeSq+nWSBwIvAB5BMw/yz2mWmP54VV07sYSSJElSj2YtyEluAzwBOKeqPgh8cOKpJEmSpIHMOga5qm4AjgEWTTyNJEmSNLBxP6R3IXDXSQaRJEmS1gTjFuR3Am9MsnCSYSRJkqShjTuLxaNpPph3UZLv0cxiUZ39VVXPnetwkiRJUt/GLcgPB24AVgD3ar+66lY/IUmSJK2Fxp3mbatJB5EkSZLWBGONQU6yaZINJh1GkiRJGtqMBTnJgiQHJbkKuAy4Oslnk9ypv3iSJElSv0YNsXgJcCBwCnAmsDXwVOBq4PkTTyZJkiQNYFRBfhHwkap68coNSV4MfDDJi6vq+omnkyRJkno2agzy1sCxU7Z9BlgA3HNiiSRJkqQBjSrIG9EMp+j6fft948nEkSRJkoY12zRvmyXZuvN8QWf7Vd0Dq+rCOU0mSZIkDWC2gnzcDNu/MM22BdNskyRJktYqowqyM1VIkiRp3pmxIFfVEX0GkSRJktYEY62kJ0mSJM0XFmRJkiSpw4IsSZIkdViQJUmSpA4LsiRJktRhQZYkSZI6LMiSJElShwVZkiRJ6rAgS5IkSR0WZEmSJKnDgixJkiR1WJAlSZKkDguyJEmS1NF7QU6ya5LzkyxL8vpp9u+c5OwkNyZ5Rt/5JEmSNL/1WpCTLAAOBXYDtgX2SrLtlMN+CTwPOKrPbJIkSRLA+j2/3g7Asqq6ECDJ0cAewHkrD6iqn7f7bu45myRJktT7EIvNgIs7z5e321ZZkv2SLEmyZMWKFXMSTpIkSeq7IGeabfWXnKiqDquqxVW1eOHChasZS5IkSWr0XZCXA1t0nm8OXNJzBkmSJGlGfRfkM4FtkmyV5LbAnsDxPWeQJEmSZtRrQa6qG4H9gZOApcAxVXVukoOT7A6QZPsky4FnAh9Ocm6fGSVJkjS/9T2LBVV1AnDClG0Hdh6fSTP0QpIkSeqdK+lJkiRJHRZkSZIkqcOCLEmSJHVYkCVJkqQOC7IkSZLUYUGWJEmSOizIkiRJUocFWZIkSeqwIEuSJEkdFmRJkiSpw4IsSZIkdViQJUmSpA4LsiRJktRhQZYkSZI6LMiSJElShwVZkiRJ6rAgS5IkSR0WZEmSJKnDgixJkiR1WJAlSZKkDguyJEmS1GFBliRJkjosyJIkSVKHBVmSJEnqsCBLkiRJHRZkSZIkqcOCLEmSJHVYkCVJkqQOC7IkSZLUYUGWJEmSOizIkiRJUocFWZIkSeqwIEuSJEkdFmRJkiSpw4IsSZIkdViQJUmSpA4LsiRJktRhQZYkSZI6LMiSJElShwVZkiRJ6rAgS5IkSR0WZEmSJKnDgixJkiR1WJAlSZKkDguyJEmS1GFBliRJkjosyJIkSVKHBVmSJEnqsCBLkiRJHRZkSZIkqcOCLEmSJHVYkCVJkqQOC7IkSZLUYUGWJEmSOizIkiRJUocFWZIkSeqwIEuSJEkdFmRJkiSpw4IsSZIkdViQJUmSpA4LsiRJktRhQZYkSZI6ei/ISXZNcn6SZUleP83+2yX5TLv/9CSL+s4oSZKk+avXgpxkAXAosBuwLbBXkm2nHLYv8NuqujfwPuAdfWaUJEnS/Nb3HeQdgGVVdWFVXQ8cDewx5Zg9gCPax8cBj0mSHjNKkiRpHktV9fdiyTOAXavqhe3zfYCHVtX+nWN+3B6zvH3+s/aY30w5137Afu3T+wDn9/AWRtkU+M2sR63bvAZeg5W8Dl4D8Bqs5HXwGoDXANaMa3DPqlo420Hr95GkY7o7wVMb+jjHUFWHAYfNRai5kGRJVS0eOseQvAZeg5W8Dl4D8Bqs5HXwGoDXANaua9D3EIvlwBad55sDl8x0TJL1gTsCV/aSTpIkSfNe3wX5TGCbJFsluS2wJ3D8lGOOB57bPn4G8M3qcxyIJEmS5rVeh1hU1Y1J9gdOAhYAH6uqc5McDCypquOBjwKfTLKM5s7xnn1mXA1rzHCPAXkNvAYreR28BuA1WMnr4DUArwGsRdeg1w/pSZIkSWs6V9KTJEmSOizIkiRJUocFeTUk2SLJyUmWJjk3ySuGzjSEJBskOSPJD9vr8OahMw0lyYIk30/y5aGzDCHJz5P8KMkPkiwZOs8QktwpyXFJftL+3bDT0Jn6luQ+7Z+BlV9XJ3nl0Ln6luRV7d+JP07y6SQbDJ2pb0le0b7/c+fTn4EkH0tyebu2w8ptmyT5WpIL2u93HjJj35I8NUklue/QWcZhQV49NwKvqar7ATsCL51m6ez54Drg0VX1AOCBwK5Jdhw401BeASwdOsTAHlVVD1xb5rqcgEOAE6vqvsADmId/Hqrq/PbPwAOBhwDXAp8fOFavkmwGvBxYXFXb0XwwfW350PmcSLId8CKaVXQfADwpyTbDpurNx4Fdp2x7PfCNqtoG+Eb7fD7ZC/g2a8n/DyzIq6GqLq2qs9vHv6f5D+Fmw6bqXzWuaZ/epv2ad5/+TLI58ETg8KGzaBhJ/g+wM81sPFTV9VV11bCpBvcY4GdV9YuhgwxgfWDDdk7/23Pref/XdfcDvldV11bVjcCpwFMHztSLqjqNW6/hsAdwRPv4COApvYYaUJKNgIcB+2JBnl+SLAIeBJw+bJJhtEMLfgBcDnytqubjdXg/8Drg5qGDDKiAryY5q10Ofr7ZGlgB/Fc71ObwJHcYOtTA9gQ+PXSIvlXVr4B3A78ELgV+V1VfHTZV734M7JzkLkluDzyBP18sbL65W1VdCs0NNuCuA+fp01NofrP2U+DKJA8eOtBsLMhzoP2X0WeBV1bV1UPnGUJV3dT+OnVzYIf2V2vzRpInAZdX1VlDZxnYw6rqwcBuNEOOdh46UM/WBx4MfKiqHgT8gfn3a9Q/aReE2h04dugsfWvHl+4BbAX8FXCHJM8eNlW/qmop8A7ga8CJwA9phiZq/tkLOLp9fHT7fI1mQV5NSW5DU46PrKrPDZ1naO2vk0/h1mOv1nUPA3ZP8nOa//M/Osmnho3Uv6q6pP1+Oc2Y0x2GTdS75cDyzm9QjqMpzPPVbsDZVXXZ0EEG8FjgoqpaUVU3AJ8D/nbgTL2rqo9W1YOrameaIQcXDJ1pQJcluQdA+/3ygfP0IsldgEcDh7f/jXwt8PdJMmiwWViQV0P7P+5HgaVV9d6h8wwlycIkd2ofb0jzH4afDJuqX1X1hqravKoW0fxK+ZtVNa/uFiW5Q5KNVz4GHk/zK9Z5o6p+DVyc5D7tpscA5w0YaWh7MQ+HV7R+CeyY5Pbtfysewzz8wGaSu7bftwSexvz98wBwPPDc9vFzgS8OmKVPzwA+UVX3rKpFVbUFcBHw8IFzjdTrUtProIcB+wA/asffAhxQVScMmGkI9wCOSLKA5h9dx1TVvJzmbJ67G/D59qbA+sBRVXXisJEG8TLgyHZ4wYXA8wfOM4h2zOnjgBcPnWUIVXV6kuOAs2mGFXyftWiZ3Tn02fYO4g3AS6vqt0MH6kOSTwOPBDZNshx4E/B24Jgk+9L8A+qZwyXs1V40773rs8DewLf6jzMel5qWJEmSOhxiIUmSJHVYkCVJkqQOC7IkSZLUYUGWJEmSOizIkiRJUocFWdKUUwnUAAAF2UlEQVS8l+R5Sarz9YckP0/y+STPSjLRvys7r79oDs/5yLk8Z5KdkhyT5JIk1ye5IsnXkjy3neJxotr3ctCkX0eSwIIsSV3PBHYCngD8G3AdzcIGX20XwZmXkrwS+A6wCfAvNIsBvQD4KfAh4EnDpZOkuedCIZJ0ix9U1bLO808mORY4FngnzSIga7T2bu6cLeGaZGfgvcAHq+rlU3Z/Mcl7gTvMwevcrqquW93zSNJc8A6yJI1QVZ+lWRL2Re3qcECzUlySdyS5qB1ycFGSN3aHYyTZIMn7kvw4yTVJfp3kS0nuO85rJ3lRkh8m+WOS3yT5aJJNphxTSd6a5PVJLgKuB/5mhvPtneT7bZbfJflRktlWuns9cCXwuhmuz8+q6pz2/AuTfDjJT5Ncm+TiJEcl2WxKjoPa3NslOSnJNcAx7b4FSd6S5NL2HKck+etxrpckzRXvIEvS7E4AngIsBk5Lsj5wErAt8O/Aj4AdaYZlbAK8pv252wEbA28BLm33/RPwvST3rapfz/SCSd7enucDwGuBzdrzbJfkb6vqps7hz6NZ1vqfgT8Al1TV9+ncSU7ycOBTnfOtB9wXuNOIDAtolsv9QlX9cdQFam0C/BF4A7AC+Kv2PXynfb9Tz/FF4KPAO4Cb220HAQfQ3LX+Ks01P36M15akOWNBlqTZ/bL9fo/2+17Aw4Fdquq0dts3kgC8Kck7quryqvod8MKVJ2kL50nAZe053jfdi7UfrHst8OaqOriz/afAt4EnA1/o/gjw+Kr63xHvYUfgqqp6ZWfbV0ccD7ApsCHwi1mOA6Cqzgde0cm7gGbs8i+B3YDPT/mRD1TVIZ3j7wy8Cjisqv55ZcYkNwFvHyeDJM0Fh1hI0uxW3omt9vuuNKXxu0nWX/lFUzhvQ1NGmx9sZsE4PclVwI00d3g3Au4z4vUeR/P385FTzn86cDWw85TjT5ylHAOcCdw5yaeSPCnJjHeOV0eSf2yHhVxD835X/uNiuvc7tTD/Dc145mOmbD96blNK0mgWZEma3Rbt90vb73cF7gncMOXrjHb/XQCSPBn4DLAU2Bt4KLA9zfCDDUa83l3b78umeY3/s/L8HZcyi6o6lWaWji1oiumKJF9Pcv8RP3YF8L8073VWSV4G/AfwdeBpwA7c8o+F6d7v1Nwr79BfNmX71OeSNFEOsZCk2T2RZmztWe3zK4CLgGfNcPzP2+97Asuq6nkrdyS5Dc1Y3VGuaL8/HvjtiP0r1TTH3EpVHQccl2QjmrHF7wBOTLJ5Vd08zfE3JjkFeNyYs0zsCXyjqlaOwSbJVqMiTXm+sjDfDTi3s/1us7yuJM0pC7IkjZDkacDuwCFVdW27+UTg6cA1VfWTET9+e5phBl37ALMtrPE1mg+tbVlVX1v11KNV1TXAl5NsDRxCc0d6xQyHvx04BXgXMHWat5UFeON2Jovb0wwB6Xr+KkQ7h2YIyrOAb3a277kK55Ck1WZBlqRbPDDJpsBtgS1pFsB4Jk1hfUPnuCNpit83krwH+GH7M/eiKdNPacv0icBTkrwP+DLwEJqSedWoEFX1syTvAD6Y5D7AqTR3sLegGZ98eFWdvCpvLMnBNHdiTwYuATZvs/ygqmYqx1TVaUleDbw3yf2Aj9OMK74z8BiaDyHuTVNuTwT+JckBNMNNHg08Y9yMVXVVe63emOT3NGO6twf2XZX3Kkmry4IsSbc4tv3+R+By4Gyau5fHVdWfhgNU1Q1J/o5mjuD9gK1o7nz+DPgKzVzEAB+hKbUvAF5M80G5J3PrD6fdSlUdkGQp8NL2q4CLgW8AF/wF7+10mkL8PpohHpfTFNB/GyPL+5OcQTPDxLtpZrf4PbCE5n19qT30YJpp415FM+b4VODvaKagG9dBNB+KfCGwf5v7yfz5kAtJmqh0/s6XJEmS5j1nsZAkSZI6LMiSJElShwVZkiRJ6rAgS5IkSR0WZEmSJKnDgixJkiR1WJAlSZKkDguyJEmS1PH/AfXD631SpTGmAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "data = 1 - (model_df.groupby(by='dealer_card').sum()['lose'] /\\\n",
    "            model_df.groupby(by='dealer_card').count()['lose'])\n",
    "\n",
    "fig, ax = plt.subplots(figsize=(10,6))\n",
    "ax = sns.barplot(x=data.index, \n",
    "                 y=data.values)\n",
    "ax.set_xlabel(\"Dealer's Card\",fontsize=16)\n",
    "ax.set_ylabel(\"Probability of Tie or Win\",fontsize=16)\n",
    "\n",
    "plt.tight_layout()\n",
    "# plt.savefig(fname='dealer_card_probs', dpi=150)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 197,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "data = 1 - (model_df.groupby(by='player_total_initial').sum()['lose'] /\\\n",
    "            model_df.groupby(by='player_total_initial').count()['lose'])\n",
    "\n",
    "fig, ax = plt.subplots(figsize=(10,6))\n",
    "ax = sns.barplot(x=data[:-1].index,\n",
    "                 y=data[:-1].values)\n",
    "ax.set_xlabel(\"Player's Hand Value\",fontsize=16)\n",
    "ax.set_ylabel(\"Probability of Tie or Win\",fontsize=16)\n",
    "\n",
    "plt.tight_layout()\n",
    "# plt.savefig(fname='player_hand_probs', dpi=150)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 198,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "has_ace\n",
       "0    0.642593\n",
       "1    0.355287\n",
       "Name: lose, dtype: float64"
      ]
     },
     "execution_count": 198,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model_df.groupby(by='has_ace').sum()['lose'] / model_df.groupby(by='has_ace').count()['lose']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 199,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(123.0, 0.5, \"Dealer's Card\")"
      ]
     },
     "execution_count": 199,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x576 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "pivot_data = model_df[model_df['player_total_initial'] != 21]\n",
    "\n",
    "losses_pivot = pd.pivot_table(pivot_data, values='lose', \n",
    "                              index=['dealer_card_num'],\n",
    "                              columns = ['player_total_initial'],\n",
    "                              aggfunc = np.sum)\n",
    "\n",
    "games_pivot =  pd.pivot_table(pivot_data, values='lose', \n",
    "                              index=['dealer_card_num'],\n",
    "                              columns = ['player_total_initial'],\n",
    "                              aggfunc = 'count')\n",
    "\n",
    "heat_data = 1 - losses_pivot.sort_index(ascending=False) / games_pivot.sort_index(ascending=False)\n",
    "\n",
    "fig, ax = plt.subplots(figsize=(16,8))\n",
    "sns.heatmap(heat_data, square=False, cmap=\"PiYG\");\n",
    "\n",
    "ax.set_xlabel(\"Player's Hand Value\",fontsize=16)\n",
    "ax.set_ylabel(\"Dealer's Card\",fontsize=16)\n",
    "\n",
    "# plt.savefig(fname='heat_map_random', dpi=150)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 200,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['player_total_initial', 'hit?', 'has_ace', 'dealer_card_num']"
      ]
     },
     "execution_count": 200,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "feature_list"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 201,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['player_total_initial', 'hit?', 'has_ace', 'dealer_card_num']\n",
      "\n",
      "\n",
      "Epoch 1/200\n",
      "257860/257860 [==============================] - 2s 7us/step - loss: 0.6949\n",
      "Epoch 2/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6204\n",
      "Epoch 3/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6166\n",
      "Epoch 4/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6152\n",
      "Epoch 5/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6143\n",
      "Epoch 6/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6141\n",
      "Epoch 7/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6142\n",
      "Epoch 8/200\n",
      "257860/257860 [==============================] - ETA: 0s - loss: 0.613 - 1s 6us/step - loss: 0.6139\n",
      "Epoch 9/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6139\n",
      "Epoch 10/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6138\n",
      "Epoch 11/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6135\n",
      "Epoch 12/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6136\n",
      "Epoch 13/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6135\n",
      "Epoch 14/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6135\n",
      "Epoch 15/200\n",
      "257860/257860 [==============================] - 2s 7us/step - loss: 0.6135\n",
      "Epoch 16/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6135A: 0s - \n",
      "Epoch 17/200\n",
      "257860/257860 [==============================] - 2s 7us/step - loss: 0.6133\n",
      "Epoch 18/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6134A\n",
      "Epoch 19/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6133A:\n",
      "Epoch 20/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6134\n",
      "Epoch 21/200\n",
      "257860/257860 [==============================] - 2s 7us/step - loss: 0.6132\n",
      "Epoch 22/200\n",
      "257860/257860 [==============================] - 2s 7us/step - loss: 0.6134A: 0s - loss\n",
      "Epoch 23/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6132\n",
      "Epoch 24/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6134\n",
      "Epoch 25/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6133\n",
      "Epoch 26/200\n",
      "257860/257860 [==============================] - 2s 7us/step - loss: 0.6132\n",
      "Epoch 27/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6133\n",
      "Epoch 28/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6132\n",
      "Epoch 29/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6132\n",
      "Epoch 30/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6133\n",
      "Epoch 31/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6131\n",
      "Epoch 32/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6132\n",
      "Epoch 33/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6131\n",
      "Epoch 34/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6132\n",
      "Epoch 35/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6132\n",
      "Epoch 36/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6131\n",
      "Epoch 37/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6131\n",
      "Epoch 38/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6131\n",
      "Epoch 39/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6133\n",
      "Epoch 40/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6132\n",
      "Epoch 41/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6130\n",
      "Epoch 42/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6131\n",
      "Epoch 43/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6131\n",
      "Epoch 44/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6131A: 0s - lo\n",
      "Epoch 45/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6131\n",
      "Epoch 46/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6131\n",
      "Epoch 47/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6130A: 0s - loss:\n",
      "Epoch 48/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6131\n",
      "Epoch 49/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6131TA: 0s - loss: 0.6\n",
      "Epoch 50/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6131\n",
      "Epoch 51/200\n",
      "257860/257860 [==============================] - 2s 7us/step - loss: 0.6131\n",
      "Epoch 52/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6131\n",
      "Epoch 53/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6130\n",
      "Epoch 54/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6130A:\n",
      "Epoch 55/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6131A: 0s \n",
      "Epoch 56/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6129A\n",
      "Epoch 57/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6131\n",
      "Epoch 58/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6131A: 0s - loss: 0.\n",
      "Epoch 59/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6130\n",
      "Epoch 60/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6130A: 0s - loss: 0.6\n",
      "Epoch 61/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6130\n",
      "Epoch 62/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6130\n",
      "Epoch 63/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6130\n",
      "Epoch 64/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6130\n",
      "Epoch 65/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6130\n",
      "Epoch 66/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6130\n",
      "Epoch 67/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6130A:\n",
      "Epoch 68/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6129\n",
      "Epoch 69/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6130\n",
      "Epoch 70/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6130\n",
      "Epoch 71/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6130\n",
      "Epoch 72/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6130\n",
      "Epoch 73/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6130A: 0s - loss: 0\n",
      "Epoch 74/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6130\n",
      "Epoch 75/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6130A: 0s - loss: 0.612\n",
      "Epoch 76/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6130\n",
      "Epoch 77/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6130\n",
      "Epoch 78/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6130\n",
      "Epoch 79/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6130\n",
      "Epoch 80/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6129\n",
      "Epoch 81/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6130\n",
      "Epoch 82/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6130\n",
      "Epoch 83/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6130\n",
      "Epoch 84/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6130\n",
      "Epoch 85/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6130A\n",
      "Epoch 86/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6130\n",
      "Epoch 87/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6129\n",
      "Epoch 88/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6130\n",
      "Epoch 89/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6130\n",
      "Epoch 90/200\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6129\n",
      "Epoch 91/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6129\n",
      "Epoch 92/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6129\n",
      "Epoch 93/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6129\n",
      "Epoch 94/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6129\n",
      "Epoch 95/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6129\n",
      "Epoch 96/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6129\n",
      "Epoch 97/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6130\n",
      "Epoch 98/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6129\n",
      "Epoch 99/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6129\n",
      "Epoch 100/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6130\n",
      "Epoch 101/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6128\n",
      "Epoch 102/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6129\n",
      "Epoch 103/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6130\n",
      "Epoch 104/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6129\n",
      "Epoch 105/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6129\n",
      "Epoch 106/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6128\n",
      "Epoch 107/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6129\n",
      "Epoch 108/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6129\n",
      "Epoch 109/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6129\n",
      "Epoch 110/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6129\n",
      "Epoch 111/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6129A: 0\n",
      "Epoch 112/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6129\n",
      "Epoch 113/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6129\n",
      "Epoch 114/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6129\n",
      "Epoch 115/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6130\n",
      "Epoch 116/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6129\n",
      "Epoch 117/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6129A: 0s - loss: 0.\n",
      "Epoch 118/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6129\n",
      "Epoch 119/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6129\n",
      "Epoch 120/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6129\n",
      "Epoch 121/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6129\n",
      "Epoch 122/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6129A: 0s\n",
      "Epoch 123/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6129\n",
      "Epoch 124/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6129\n",
      "Epoch 125/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6129\n",
      "Epoch 126/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6129\n",
      "Epoch 127/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6129\n",
      "Epoch 128/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6129\n",
      "Epoch 129/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6129\n",
      "Epoch 130/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6129\n",
      "Epoch 131/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6129\n",
      "Epoch 132/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6129\n",
      "Epoch 133/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6128\n",
      "Epoch 134/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6128A: 0s - loss: 0.61\n",
      "Epoch 135/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6129\n",
      "Epoch 136/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6129A: 0s - loss: \n",
      "Epoch 137/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6129\n",
      "Epoch 138/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6129\n",
      "Epoch 139/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6129\n",
      "Epoch 140/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6129\n",
      "Epoch 141/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6128\n",
      "Epoch 142/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6128\n",
      "Epoch 143/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6130\n",
      "Epoch 144/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6129\n",
      "Epoch 145/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6129\n",
      "Epoch 146/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6129A: 0s - loss: 0.61 - ETA: 0s - loss:\n",
      "Epoch 147/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6129\n",
      "Epoch 148/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6129\n",
      "Epoch 149/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6129\n",
      "Epoch 150/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6128\n",
      "Epoch 151/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6128\n",
      "Epoch 152/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6128\n",
      "Epoch 153/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6129\n",
      "Epoch 154/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6129\n",
      "Epoch 155/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6128\n",
      "Epoch 156/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6129\n",
      "Epoch 157/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6128\n",
      "Epoch 158/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6128\n",
      "Epoch 159/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6129\n",
      "Epoch 160/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6129\n",
      "Epoch 161/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6129\n",
      "Epoch 162/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6129\n",
      "Epoch 163/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6129\n",
      "Epoch 164/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6128\n",
      "Epoch 165/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6128A: 0s - \n",
      "Epoch 166/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6128A: 0s - loss: \n",
      "Epoch 167/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6129\n",
      "Epoch 168/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6128\n",
      "Epoch 169/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6128\n",
      "Epoch 170/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6128\n",
      "Epoch 171/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6128\n",
      "Epoch 172/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6128\n",
      "Epoch 173/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6128\n",
      "Epoch 174/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6128\n",
      "Epoch 175/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6128\n",
      "Epoch 176/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6128\n",
      "Epoch 177/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6129\n",
      "Epoch 178/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6128\n",
      "Epoch 179/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6128\n",
      "Epoch 180/200\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6129A: 0s - loss: \n",
      "Epoch 181/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6128\n",
      "Epoch 182/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6128A: 0s - loss: 0.6\n",
      "Epoch 183/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6128\n",
      "Epoch 184/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6128\n",
      "Epoch 185/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6128\n",
      "Epoch 186/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6128\n",
      "Epoch 187/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6128A: 0s -\n",
      "Epoch 188/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6128\n",
      "Epoch 189/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6128\n",
      "Epoch 190/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.6128A: 0s - loss: 0.\n",
      "Epoch 191/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6128\n",
      "Epoch 192/200\n",
      "257860/257860 [==============================] - 2s 8us/step - loss: 0.6128\n",
      "Epoch 193/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6128\n",
      "Epoch 194/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6128\n",
      "Epoch 195/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6128\n",
      "Epoch 196/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6128\n",
      "Epoch 197/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6129A: 0s - lo - ETA: 0s - loss: 0.61\n",
      "Epoch 198/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6128\n",
      "Epoch 199/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.6128A: 0s -\n",
      "Epoch 200/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.6128A: 0s \n"
     ]
    }
   ],
   "source": [
    "# Train a neural net to play blackjack\n",
    "\n",
    "# Set up variables for neural net\n",
    "feature_list = [i for i in model_df.columns if i not in ['dealer_card',\n",
    "                                                         'Y','lose',\n",
    "                                                         'correct_action',\n",
    "                                                         'dealer_bust',\n",
    "                                                         'dealer_bust_pred',\n",
    "                                                         'new_stack', 'games_played_with_stack',\n",
    "                                                         2,3,4,5,6,7,8,9,10,'A',\n",
    "                                                         'blackjack?'\n",
    "                                                        ]]\n",
    "train_X = np.array(model_df[feature_list])\n",
    "train_Y = np.array(model_df['correct_action']).reshape(-1,1)\n",
    "\n",
    "print(feature_list)\n",
    "print('\\n')\n",
    "\n",
    "# Set up a neural net with 5 layers\n",
    "model = Sequential()\n",
    "# model.add(Dense(train_X.shape[1]+1))\n",
    "model.add(Dense(16))\n",
    "model.add(Dense(128))\n",
    "model.add(Dense(32))\n",
    "model.add(Dense(8))\n",
    "model.add(Dense(1, activation='sigmoid'))\n",
    "model.compile(loss='binary_crossentropy', optimizer='sgd')\n",
    "model.fit(train_X, train_Y, epochs=200, batch_size=256, verbose=1)\n",
    "\n",
    "pred_Y_train = model.predict(train_X)\n",
    "actuals = train_Y[:,-1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 218,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot ROC Curve\n",
    "\n",
    "fpr, tpr, threshold = metrics.roc_curve(actuals, pred_Y_train)\n",
    "roc_auc = metrics.auc(fpr, tpr)\n",
    "\n",
    "fig, ax = plt.subplots(figsize=(10,8))\n",
    "plt.plot(fpr, tpr, label = ('ROC AUC = %0.3f' % roc_auc))\n",
    "\n",
    "plt.legend(loc = 'lower right')\n",
    "plt.plot([0, 1], [0, 1],'r--')\n",
    "plt.xlim([0, 1])\n",
    "plt.ylim([0, 1])\n",
    "ax.set_xlabel(\"False Positive Rate\",fontsize=16)\n",
    "ax.set_ylabel(\"True Positive Rate\",fontsize=16)\n",
    "plt.setp(ax.get_legend().get_texts(), fontsize=16)\n",
    "plt.tight_layout()\n",
    "plt.savefig(fname='roc_curve_blackjack', dpi=150)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 219,
   "metadata": {},
   "outputs": [],
   "source": [
    "def func(x):\n",
    "    if x == 21:\n",
    "        return 1\n",
    "    else:\n",
    "        return 0\n",
    "\n",
    "model_df['blackjack?'] = model_df['player_total_initial'].apply(func)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 204,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[2, 3, 4, 5, 6, 7, 8, 9, 10, 'A', 'dealer_card_num']\n",
      "\n",
      "\n",
      "Epoch 1/200\n",
      "257860/257860 [==============================] - 2s 7us/step - loss: 0.2010\n",
      "Epoch 2/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.1857\n",
      "Epoch 3/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1816\n",
      "Epoch 4/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1795\n",
      "Epoch 5/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1781\n",
      "Epoch 6/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1774\n",
      "Epoch 7/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.1769\n",
      "Epoch 8/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1762\n",
      "Epoch 9/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1751A: 0s - loss: 0.\n",
      "Epoch 10/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1745\n",
      "Epoch 11/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1740\n",
      "Epoch 12/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1735\n",
      "Epoch 13/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.1731\n",
      "Epoch 14/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1726\n",
      "Epoch 15/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1723\n",
      "Epoch 16/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1720\n",
      "Epoch 17/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1718\n",
      "Epoch 18/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1715\n",
      "Epoch 19/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1713\n",
      "Epoch 20/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1711A: 0s - loss: 0.171\n",
      "Epoch 21/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1710\n",
      "Epoch 22/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1709\n",
      "Epoch 23/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.1707A: 0s - loss: 0.\n",
      "Epoch 24/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1707\n",
      "Epoch 25/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1706\n",
      "Epoch 26/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1705\n",
      "Epoch 27/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1705A: 0s - \n",
      "Epoch 28/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1704A: 0s - \n",
      "Epoch 29/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1703\n",
      "Epoch 30/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.1702A: 0\n",
      "Epoch 31/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1702\n",
      "Epoch 32/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1702A: 0s - \n",
      "Epoch 33/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1701\n",
      "Epoch 34/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1701A: \n",
      "Epoch 35/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1700\n",
      "Epoch 36/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1700\n",
      "Epoch 37/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.1700\n",
      "Epoch 38/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1699\n",
      "Epoch 39/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1698\n",
      "Epoch 40/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1699\n",
      "Epoch 41/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1698\n",
      "Epoch 42/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1697\n",
      "Epoch 43/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.1697\n",
      "Epoch 44/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1697A: 0s - los\n",
      "Epoch 45/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1697A: 0s - \n",
      "Epoch 46/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1696A: 0s - los\n",
      "Epoch 47/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1697A: 0s - \n",
      "Epoch 48/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1696\n",
      "Epoch 49/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1695\n",
      "Epoch 50/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1695\n",
      "Epoch 51/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1695A:\n",
      "Epoch 52/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1695\n",
      "Epoch 53/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1694\n",
      "Epoch 54/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1694\n",
      "Epoch 55/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1694\n",
      "Epoch 56/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1693\n",
      "Epoch 57/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1692\n",
      "Epoch 58/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1692\n",
      "Epoch 59/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1692\n",
      "Epoch 60/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1692\n",
      "Epoch 61/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1691\n",
      "Epoch 62/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1691\n",
      "Epoch 63/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1691\n",
      "Epoch 64/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1691\n",
      "Epoch 65/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1691\n",
      "Epoch 66/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1690\n",
      "Epoch 67/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1690\n",
      "Epoch 68/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1689\n",
      "Epoch 69/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1689\n",
      "Epoch 70/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1689\n",
      "Epoch 71/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1688\n",
      "Epoch 72/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1689\n",
      "Epoch 73/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1688\n",
      "Epoch 74/200\n",
      "257860/257860 [==============================] - ETA: 0s - loss: 0.168 - 1s 5us/step - loss: 0.1688\n",
      "Epoch 75/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1687\n",
      "Epoch 76/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1687\n",
      "Epoch 77/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1687\n",
      "Epoch 78/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1686\n",
      "Epoch 79/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1686\n",
      "Epoch 80/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1686A: 0s - loss: 0.169 - ETA: 0s \n",
      "Epoch 81/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1685\n",
      "Epoch 82/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1685\n",
      "Epoch 83/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1685\n",
      "Epoch 84/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1684\n",
      "Epoch 85/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1684\n",
      "Epoch 86/200\n",
      "257860/257860 [==============================] - 2s 7us/step - loss: 0.1684\n",
      "Epoch 87/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1684A: 0s - loss: 0.168\n",
      "Epoch 88/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.1684\n",
      "Epoch 89/200\n",
      "257860/257860 [==============================] - 2s 7us/step - loss: 0.1683A: 0s - \n",
      "Epoch 90/200\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.1683A: 0s - loss\n",
      "Epoch 91/200\n",
      "257860/257860 [==============================] - 2s 7us/step - loss: 0.1682\n",
      "Epoch 92/200\n",
      "257860/257860 [==============================] - 2s 7us/step - loss: 0.1683A: 0s - \n",
      "Epoch 93/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.1681\n",
      "Epoch 94/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.1682A: 0s - loss\n",
      "Epoch 95/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.1682A: 0s - loss: 0.1 - ETA: 0s - loss: 0.1\n",
      "Epoch 96/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.1682\n",
      "Epoch 97/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1681\n",
      "Epoch 98/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.1681\n",
      "Epoch 99/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.1681\n",
      "Epoch 100/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1680\n",
      "Epoch 101/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1680\n",
      "Epoch 102/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1681\n",
      "Epoch 103/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1680\n",
      "Epoch 104/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1679\n",
      "Epoch 105/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1680\n",
      "Epoch 106/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1680A: 0s - los\n",
      "Epoch 107/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1679\n",
      "Epoch 108/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1679\n",
      "Epoch 109/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1679\n",
      "Epoch 110/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1678\n",
      "Epoch 111/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1679\n",
      "Epoch 112/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.1678\n",
      "Epoch 113/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1678\n",
      "Epoch 114/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1677\n",
      "Epoch 115/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1678\n",
      "Epoch 116/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1677\n",
      "Epoch 117/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1677\n",
      "Epoch 118/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1677\n",
      "Epoch 119/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1677\n",
      "Epoch 120/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.1677\n",
      "Epoch 121/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1677\n",
      "Epoch 122/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1677\n",
      "Epoch 123/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1677\n",
      "Epoch 124/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1677\n",
      "Epoch 125/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1676\n",
      "Epoch 126/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1676\n",
      "Epoch 127/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1676\n",
      "Epoch 128/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1676\n",
      "Epoch 129/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1675A:\n",
      "Epoch 130/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1675A: 0s \n",
      "Epoch 131/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1675\n",
      "Epoch 132/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1676\n",
      "Epoch 133/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1675\n",
      "Epoch 134/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.1675\n",
      "Epoch 135/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1675\n",
      "Epoch 136/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1675\n",
      "Epoch 137/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1675\n",
      "Epoch 138/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1675\n",
      "Epoch 139/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1674A:\n",
      "Epoch 140/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1675\n",
      "Epoch 141/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.1674\n",
      "Epoch 142/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1674\n",
      "Epoch 143/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.1673\n",
      "Epoch 144/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.1675\n",
      "Epoch 145/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1674\n",
      "Epoch 146/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1674\n",
      "Epoch 147/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1673\n",
      "Epoch 148/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.1674\n",
      "Epoch 149/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1674\n",
      "Epoch 150/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1673\n",
      "Epoch 151/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1673\n",
      "Epoch 152/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1673\n",
      "Epoch 153/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1673\n",
      "Epoch 154/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1673\n",
      "Epoch 155/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1672\n",
      "Epoch 156/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1672\n",
      "Epoch 157/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1673\n",
      "Epoch 158/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1672\n",
      "Epoch 159/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1672\n",
      "Epoch 160/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1672A: 0s - loss: \n",
      "Epoch 161/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1672\n",
      "Epoch 162/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1672\n",
      "Epoch 163/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1672\n",
      "Epoch 164/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1671\n",
      "Epoch 165/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1672\n",
      "Epoch 166/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1671\n",
      "Epoch 167/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1672\n",
      "Epoch 168/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.1671\n",
      "Epoch 169/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1671\n",
      "Epoch 170/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1672\n",
      "Epoch 171/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1672\n",
      "Epoch 172/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1671\n",
      "Epoch 173/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1671A: \n",
      "Epoch 174/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1671\n",
      "Epoch 175/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1671\n",
      "Epoch 176/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1670\n",
      "Epoch 177/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1670\n",
      "Epoch 178/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1670\n",
      "Epoch 179/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1671\n",
      "Epoch 180/200\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1671\n",
      "Epoch 181/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1670\n",
      "Epoch 182/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1670\n",
      "Epoch 183/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1670\n",
      "Epoch 184/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1670\n",
      "Epoch 185/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1669\n",
      "Epoch 186/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1670A: 0s - loss: \n",
      "Epoch 187/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1670\n",
      "Epoch 188/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1669\n",
      "Epoch 189/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1669\n",
      "Epoch 190/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1670\n",
      "Epoch 191/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1669\n",
      "Epoch 192/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1669\n",
      "Epoch 193/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1669A: \n",
      "Epoch 194/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1669\n",
      "Epoch 195/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1669\n",
      "Epoch 196/200\n",
      "257860/257860 [==============================] - 1s 6us/step - loss: 0.1669\n",
      "Epoch 197/200\n",
      "257860/257860 [==============================] - 2s 6us/step - loss: 0.1668\n",
      "Epoch 198/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1668\n",
      "Epoch 199/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1669\n",
      "Epoch 200/200\n",
      "257860/257860 [==============================] - 1s 5us/step - loss: 0.1668\n"
     ]
    }
   ],
   "source": [
    "# Train a neural net to PREDICT BLACKJACKS\n",
    "# Apologize for the name, it started as a model to predict dealer busts\n",
    "# Then I decided to predict blackjacks instead but neglected to rename it\n",
    "\n",
    "bust_features = [2,3,4,5,6,7,8,9,10,'A','dealer_card_num']\n",
    "\n",
    "# Set up variables for neural net\n",
    "feature_list_bust = [i for i in bust_features if i not in ['dealer_bust']]\n",
    "train_X_bust = np.array(model_df[feature_list_bust])\n",
    "train_Y_bust = np.array(model_df['blackjack?']).reshape(-1,1)\n",
    "\n",
    "print(feature_list_bust)\n",
    "print('\\n')\n",
    "\n",
    "# Set up a neural net with 5 layers\n",
    "model_bust = Sequential()\n",
    "model_bust.add(Dense(train_X_bust.shape[1]))\n",
    "model_bust.add(Dense(128))\n",
    "model_bust.add(Dense(32, activation='relu'))\n",
    "model_bust.add(Dense(8))\n",
    "model_bust.add(Dense(1, activation='sigmoid'))\n",
    "model_bust.compile(loss='binary_crossentropy', optimizer='sgd')\n",
    "model_bust.fit(train_X_bust, train_Y_bust, epochs=200, batch_size=256, verbose=1)\n",
    "\n",
    "pred_Y_train_bust = model_bust.predict(train_X_bust)\n",
    "actuals = train_Y_bust[:,-1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 220,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot ROC Curve for predicting dealer bust\n",
    "\n",
    "fpr, tpr, threshold = metrics.roc_curve(actuals, pred_Y_train_bust)\n",
    "roc_auc = metrics.auc(fpr, tpr)\n",
    "\n",
    "fig, ax = plt.subplots(figsize=(10,8))\n",
    "plt.plot(fpr, tpr, label = ('ROC AUC = %0.3f' % roc_auc))\n",
    "\n",
    "plt.legend(loc = 'lower right')\n",
    "plt.plot([0, 1], [0, 1],'r--')\n",
    "plt.xlim([0, 1])\n",
    "plt.ylim([0, 1])\n",
    "ax.set_xlabel(\"False Positive Rate\",fontsize=16)\n",
    "ax.set_ylabel(\"True Positive Rate\",fontsize=16)\n",
    "plt.setp(ax.get_legend().get_texts(), fontsize=16)\n",
    "plt.tight_layout()\n",
    "plt.savefig(fname='roc_curve_blackjack', dpi=150)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 206,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Given the relevant inputs, the function below uses the neural net to make a prediction\n",
    "# and then based on that prediction, decides whether to hit or stay\n",
    "\n",
    "def model_decision(model, player_sum, has_ace, dealer_card_num, new_stack, games_played, card_count):\n",
    "    input_array = np.array([player_sum, 0, has_ace, \n",
    "                            dealer_card_num, new_stack, \n",
    "                            games_played]).reshape(1,-1)\n",
    "    cc_array = pd.DataFrame.from_dict([card_count])\n",
    "    input_array = np.concatenate([input_array, cc_array], axis=1)\n",
    "    predict_correct = model.predict(input_array)\n",
    "    if predict_correct >= 0.52:\n",
    "        return 1, predict_correct\n",
    "    else:\n",
    "        return 0, predict_correct\n",
    "    \n",
    "def model_decision_old(model, player_sum, has_ace, dealer_card_num):\n",
    "    input_array = np.array([player_sum, 0, has_ace, dealer_card_num]).reshape(1,-1)\n",
    "    predict_correct = model.predict(input_array)\n",
    "    if predict_correct >= 0.52:\n",
    "        return 1\n",
    "    else:\n",
    "        return 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 207,
   "metadata": {},
   "outputs": [],
   "source": [
    "def bust_Z_score(pred, pred_mean, pred_std):\n",
    "    return (pred - pred_mean)/pred_std\n",
    "\n",
    "pred_mean = model_bust.predict(train_X_bust).mean()\n",
    "pred_std = model_bust.predict(train_X_bust).std()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 269,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Play blackjack but use the neural net to make hit/stay decision\n",
    "# And use a second neural net to decide how much to bet based on probability of dealer busting\n",
    "\n",
    "nights = 101\n",
    "bankrolls = []\n",
    "\n",
    "for night in range(nights):\n",
    "    \n",
    "    dollars = 10000\n",
    "    bankroll = []\n",
    "    stacks = 101\n",
    "    players = 1\n",
    "    num_decks = 1\n",
    "\n",
    "    card_types = ['A',2,3,4,5,6,7,8,9,10,10,10,10]\n",
    "\n",
    "    dealer_card_feature = []\n",
    "    player_card_feature = []\n",
    "    player_live_total = []\n",
    "    player_live_action = []\n",
    "    player_results = []\n",
    "\n",
    "    first_game = True\n",
    "    prev_stack = 0\n",
    "    stack_num_list = []\n",
    "    new_stack = []\n",
    "    card_count_list = []\n",
    "    games_played_with_stack = []\n",
    "    \n",
    "    for stack in range(stacks):\n",
    "        games_played = 0\n",
    "\n",
    "        if stack != prev_stack:\n",
    "            temp_new_stack = 1\n",
    "        else:\n",
    "            temp_new_stack = 0\n",
    "\n",
    "        # Make a dict for keeping track of the count for a stack\n",
    "        card_count = {2: 0,\n",
    "                      3: 0,\n",
    "                      4: 0,\n",
    "                      5: 0,\n",
    "                      6: 0,\n",
    "                      7: 0,\n",
    "                      8: 0,\n",
    "                      9: 0,\n",
    "                      10: 0,\n",
    "                      'A': 0}\n",
    "\n",
    "        blackjack = set(['A',10])\n",
    "        dealer_cards = make_decks(num_decks, card_types)\n",
    "        while len(dealer_cards) > 20:\n",
    "            multiplier = 1\n",
    "\n",
    "            curr_player_results = np.zeros((1,players))\n",
    "\n",
    "            dealer_hand = []\n",
    "            player_hands = [[] for player in range(players)]\n",
    "            live_total = []\n",
    "            live_action = []\n",
    "\n",
    "            # Record card count\n",
    "            cc_array_bust = pd.DataFrame.from_dict([card_count])\n",
    "\n",
    "            # Deal FIRST card\n",
    "            for player, hand in enumerate(player_hands):\n",
    "                player_hands[player].append(dealer_cards.pop(0))\n",
    "            dealer_hand.append(dealer_cards.pop(0))\n",
    "            # Deal SECOND card\n",
    "            for player, hand in enumerate(player_hands):\n",
    "                player_hands[player].append(dealer_cards.pop(0))\n",
    "            dealer_hand.append(dealer_cards.pop(0))\n",
    "\n",
    "            # Record the player's live total after cards are dealt\n",
    "            live_total.append(total_up(player_hands[player]))\n",
    "            action = 0\n",
    "\n",
    "            # Dealer checks for 21\n",
    "            if set(dealer_hand) == blackjack:\n",
    "                for player in range(players):\n",
    "                    if set(player_hands[player]) != blackjack:\n",
    "                        curr_player_results[0,player] = -1\n",
    "                    else:\n",
    "                        curr_player_results[0,player] = 0\n",
    "            else:\n",
    "                for player in range(players):\n",
    "                    # Players check for 21\n",
    "                    if set(player_hands[player]) == blackjack:\n",
    "                        curr_player_results[0,player] = 1\n",
    "                        multiplier = 1.25\n",
    "                    else:\n",
    "                        # Neural net decides whether to hit or stay\n",
    "                        if 'A' in player_hands[player][0:2]:\n",
    "                            ace_in_hand = 1\n",
    "                        else:\n",
    "                            ace_in_hand = 0\n",
    "                        if dealer_hand[0] == 'A':\n",
    "                            dealer_face_up_card = 11\n",
    "                        else:\n",
    "                            dealer_face_up_card = dealer_hand[0]\n",
    "\n",
    "                        while (model_decision_old(model, total_up(player_hands[player]), \n",
    "                                                  ace_in_hand, dealer_face_up_card,\n",
    "                                                  #temp_new_stack, games_played, \n",
    "                                                  #card_count\n",
    "                                                 ) == 1) and (total_up(player_hands[player]) != 21):\n",
    "                            player_hands[player].append(dealer_cards.pop(0))\n",
    "                            action = 1\n",
    "                            live_total.append(total_up(player_hands[player]))\n",
    "                            if total_up(player_hands[player]) > 21:\n",
    "                                curr_player_results[0,player] = -1\n",
    "                                break\n",
    "\n",
    "            # Dealer hits based on the rules\n",
    "            while total_up(dealer_hand) < 17:\n",
    "                dealer_hand.append(dealer_cards.pop(0))\n",
    "            # Compare dealer hand to players hand but first check if dealer busted\n",
    "            if total_up(dealer_hand) > 21:\n",
    "                for player in range(players):\n",
    "                    if curr_player_results[0,player] != -1:\n",
    "                        curr_player_results[0,player] = 1\n",
    "            else:\n",
    "                for player in range(players):\n",
    "                    if total_up(player_hands[player]) > total_up(dealer_hand):\n",
    "                        if total_up(player_hands[player]) <= 21:\n",
    "                            curr_player_results[0,player] = 1\n",
    "                    elif total_up(player_hands[player]) == total_up(dealer_hand):\n",
    "                        curr_player_results[0,player] = 0\n",
    "                    else:\n",
    "                        curr_player_results[0,player] = -1\n",
    "\n",
    "            input_array = np.concatenate([cc_array_bust, \n",
    "                                          np.array(dealer_face_up_card).reshape(1,-1)], axis=1)\n",
    "            bust_pred = model_bust.predict(input_array)\n",
    "            bust_Z = bust_Z_score(bust_pred, pred_mean, pred_std)\n",
    "\n",
    "            if bust_Z >= 0:\n",
    "                bet = 100#*(1 + bust_Z_score)\n",
    "            else:\n",
    "                bet = 100\n",
    "            dollars+=curr_player_results[0,player]*bet*multiplier\n",
    "            bankroll.append(dollars)\n",
    "\n",
    "            # Track features\n",
    "            dealer_card_feature.append(dealer_hand[0])\n",
    "            player_card_feature.append(player_hands)\n",
    "            player_results.append(list(curr_player_results[0]))\n",
    "            player_live_total.append(live_total)\n",
    "            player_live_action.append(action)\n",
    "\n",
    "            # Update card count list with most recent game's card count\n",
    "            if stack != prev_stack:\n",
    "                new_stack.append(1)\n",
    "            else:\n",
    "                new_stack.append(0)\n",
    "                if first_game == True:\n",
    "                    first_game = False\n",
    "                else:\n",
    "                    games_played += 1\n",
    "\n",
    "            stack_num_list.append(stack)\n",
    "            games_played_with_stack.append(games_played)\n",
    "            card_count_list.append(card_count.copy())\n",
    "            prev_stack = stack\n",
    "\n",
    "    model_df_smart = pd.DataFrame()\n",
    "    model_df_smart['dealer_card'] = dealer_card_feature\n",
    "    model_df_smart['player_total_initial'] = [total_up(i[0][0:2]) for i in player_card_feature]\n",
    "    model_df_smart['hit?'] = player_live_action\n",
    "\n",
    "    has_ace = []\n",
    "    for i in player_card_feature:\n",
    "        if ('A' in i[0][0:2]):\n",
    "            has_ace.append(1)\n",
    "        else:\n",
    "            has_ace.append(0)\n",
    "    model_df_smart['has_ace'] = has_ace\n",
    "\n",
    "    dealer_card_num = []\n",
    "    for i in model_df_smart['dealer_card']:\n",
    "        if i=='A':\n",
    "            dealer_card_num.append(11)\n",
    "        else:\n",
    "            dealer_card_num.append(i)\n",
    "    model_df_smart['dealer_card_num'] = dealer_card_num\n",
    "\n",
    "    model_df_smart['Y'] = [i[0] for i in player_results]\n",
    "    lose = []\n",
    "    for i in model_df_smart['Y']:\n",
    "        if i == -1:\n",
    "            lose.append(1)\n",
    "        else:\n",
    "            lose.append(0)\n",
    "    model_df_smart['lose'] = lose\n",
    "    \n",
    "    bankrolls.append(bankroll)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 271,
   "metadata": {},
   "outputs": [],
   "source": [
    "# bet_results = pd.DataFrame(bankrolls).T.fillna(method='ffill',axis=0)\n",
    "# bet_results.to_csv('dumb_bet.csv')\n",
    "# static_smart_bets = pd.read_csv('static_smart_bet.csv')\n",
    "smart_bets = pd.read_csv('smart_bet.csv')\n",
    "dumb_bets = pd.read_csv('dumb_bet.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 274,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAsgAAAGoCAYAAABbtxOxAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvOIA7rQAAIABJREFUeJzs3Xt8zuX/wPHXtbPNbGYOY5jDHGeGOS3kkGNOCZGwJBUi/VL6qiypr6JvCoUIIRQJSTmEnG2Y05wZ5mxjdrDDvV2/Pz63O6fZ2ObeeD8fjz3mvj7X5/q87xv1du19XZfSWiOEEEIIIYQw2Fg7ACGEEEIIIfISSZCFEEIIIYS4hSTIQgghhBBC3EISZCGEEEIIIW4hCbIQQgghhBC3kARZCCGEEEKIW0iCLIQQQgghxC0kQRZCCCGEEOIWkiALIYQQQghxCztrB5AXeHp6ah8fH2uHIYQQQgghctHOnTuvaK2LZtZPEmTAx8eHsLAwa4chhBBCCCFykVLqVFb6SYmFEEIIIYQQt5AEWQghhBBCiFtIgiyEEEIIIcQtpAZZCCGEuI/U1FSioqJISkqydihCiCxycnLC29sbe3v7h7r/kSbISqkfgPbAJa21n7nNA1gI+ACRQHet9VWllAK+BtoBiUCw1nqX+Z6+wAfmYcdorWeb2+sAs4ACwB/AUK21fiRvTgghxGMpKioKV1dXfHx8MP7XJITIy7TWREdHExUVRbly5R5qjEddYjELaHNH2whgrdbaF1hrfg3QFvA1fw0AvgNLQj0KqA/UA0YppQqb7/nO3PfmfXc+SwghhHggSUlJFClSRJJjIfIJpRRFihTJ1k99HmmCrLX+B4i5o7kTMNv869lA51vaf9SGbYC7UsoLaA2s1lrHaK2vAquBNuZrhbTWW82zxj/eMpYQQgjx0CQ5FiJ/ye7f2bywSK+41vo8gPl7MXN7KeDMLf2izG33a4+6R/s9KaUGKKXClFJhly9fzvabEEIIIYQQj4e8kCBn5F6pv36I9nvSWk/TWgdqrQOLFs30QBUhhBDCapRS9O7d2/LaZDJRtGhR2rdvn6vPDQ4Oply5cgQEBFClShU+/vjjhx5r1qxZDB48OEfuDwkJYfz48Q90f9OmTQkMDLS8DgsLo2nTpg8dT1ZlFGtISAjOzs5cunTJ0lawYMFMx/vss8/u2Z6YmMizzz5LlSpVqF69OiNGjLhnv4sXL9K+fXtq1qxJtWrVaNeuHQDnzp2ja9euGT43s+sPYtasWRQtWpSAgACqV69O165dSUxMvO8969evZ8uWLTny/KzICwnyRXN5BObvN/+kRAGlb+nnDZzLpN37Hu1CCCFEvubi4sL+/fu5ceMGAKtXr6ZUqQx/SJqjxo0bR3h4OOHh4cyePZuTJ0/m2rNMJlOujQ1w6dIlVq5cmePjpqWlPdR9np6efPnllw90T0YJMsA777zDoUOH2L17N5s3b77ne/3oo49o2bIle/bsISIigrFjxwJQsmRJFi1alOHYmV1/UC+88ALh4eEcOHAABwcHFi5ceN/+jzpBzgvbvC0D+gJjzd+X3tI+WCm1AGNBXqzW+rxS6i/gs1sW5rUC3tdaxyil4pRSDYDtQB9g4qN8I0IIIR5vHy8/QMS56zk6ZrWShRjVoXqm/dq2bcuKFSvo2rUr8+fPp2fPnmzcuBGAhIQE3nzzTfbt24fJZCIkJIROnToRGRlJ7969SUhIAGDSpEkEBQWxfv16QkJC8PT0ZP/+/dSpU4e5c+fet27z5oInFxcXAEaPHs3y5cu5ceMGQUFBTJ06FaUUTZs2pX79+qxbt45r164xY8YMGjdufNtYK1asYMyYMSxfvpx33nkHDw8Pdu/eTe3atRk5ciT9+vXjxIkTODs7M23aNPz9/R/qs73T8OHDGTNmDG3btr2tPS0tjREjRrB+/XqSk5MZNGgQr732GuvXr2f8+PH8/vvvAAwePJjAwECCg4Px8fGhX79+rFq1isGDBxMXF8e0adNISUmhYsWKzJkzB2dn5/vG069fP2bNmsV7772Hh4fHbdfmzp3LN998Q0pKCvXr1+fbb79l5MiR3LhxwzLzOm/ePEt/Z2dnmjVrBoCDgwO1a9cmKiqKO50/f55WrVpZXt/8bCMjI2nfvj379++nf//+hIWFAXD27FkGDx5M3759LddnzZrFsmXLSExM5Pjx4zz33HN88cUXAMyYMYPPP/+ckiVL4uvri6OjI5MmTcrwMzCZTCQkJFC4sJHWXb58mddff53Tp08DMGHCBEqVKsWUKVOwtbVl7ty5TJw4kQsXLvDxxx9ja2uLm5sb//zzz30/6wf1SGeQlVLzga1AZaVUlFLqFYzEuKVS6ijQ0vwajG3aTgDHgO+BgQBa6xjgEyDU/DXa3AbwBjDdfM9xIOf/mSiEEEJYQY8ePViwYAFJSUns3buX+vXrW659+umnNG/enNDQUNatW8fw4cNJSEigWLFirF69ml27drFw4UKGDBliuWf37t1MmDCBiIgITpw4webNm+/53OHDhxMQEIC3tzc9evSgWDFjqdDgwYMJDQ21zGzfTCLBSHp27NjBhAkT7irLWLJkCWPHjuWPP/7A09MTgCNHjrBmzRq+/PJLRo0aRa1atdi7dy+fffYZffr0ybHPsGHDhjg6OrJu3brb2mfMmIGbmxuhoaGEhoby/fffZ2mm3MnJiU2bNtGjRw+6dOlCaGgoe/bsoWrVqsyYMSPT+wsWLEi/fv34+uuvb2s/ePAgCxcuZPPmzYSHh2Nra8u8efMYO3YsBQoUIDw8/Lbk+E7Xrl1j+fLltGjR4q5rgwYN4pVXXqFZs2Z8+umnnDt39w/bp0+fTnh4OEuXLqVIkSIEBwff1Sc8PJyFCxeyb98+Fi5cyJkzZzh37hyffPIJ27ZtY/Xq1Rw6dCjDGBcuXEhAQAClSpUiJiaGDh06ADB06FCGDRtGaGgoixcvpn///vj4+PD6668zbNgwwsPDady4MaNHj+avv/5iz549LFu2LMPnPKxHOoOste6ZwaW7fgfNO1EMymCcH4Af7tEeBvhlJ0YhhBAiI1mZ6c0t/v7+REZGMn/+fEvd6E2rVq1i2bJlllrXpKQkTp8+TcmSJRk8eLAlyTpy5Ijlnnr16uHtbVQmBgQEEBkZSaNGje567rhx4+jatSvx8fG0aNGCLVu2EBQUxLp16/jiiy9ITEwkJiaG6tWrW5KcLl26AFCnTh0iIyMtY61bt46wsDBWrVpFoUKFLO3dunXD1tYWgE2bNrF48WIAmjdvTnR0NLGxsdn9+Cw++OADxowZw+eff25pW7VqFXv37rWUEMTGxnL06FEcHBzuO9YLL7xg+fX+/fv54IMPuHbtGvHx8bRu3TpL8QwZMoSAgAD+7//+z9K2du1adu7cSd26dQG4ceOG5R8mmTGZTPTs2ZMhQ4ZQvnz5u663bt2aEydO8Oeff7Jy5Upq1arF/v377+qXlJREt27dmDRpEmXLlr3t9xGgRYsWuLm5AVCtWjVOnTrFlStXePrppy2z4d26dbvtz9ytXnjhBSZNmoTWmkGDBjFu3DhGjBjBmjVriIiIsPS7fv06cXFxd93/1FNPERwcTPfu3S1/3nJSXqhBfvLEy64ZQgghHlzHjh1555136Nnz9vkmrTWLFy+21AqfPn2aqlWr8tVXX1G8eHH27NlDWFgYKSkplnscHR0tv7a1tc20/rdgwYI0bdqUTZs2kZSUxMCBA1m0aBH79u3j1VdfvW3P2Ztj3zlu+fLliYuLuytpulm2cfO93CkrW3alpaUREBBAQEAAH330UYb9mjdvTlJSEtu2bbvtmRMnTrR8fidPnqRVq1bY2dmRnp5u6Xfnvrq3xh0cHMykSZPYt28fo0aNyvIevO7u7rz44ot8++23t8XTt29fSzyHDx8mJCQkS+MNGDAAX19f3nrrrQz7eHh48OKLLzJnzhzq1q17z/KE119/nS5duvDMM8/cc4x7/fl5mLPZlFJ06NDBEkN6ejpbt261vPezZ8/i6up6131TpkxhzJgxnDlzhoCAAKKjox/42fcjCfKjdi6c9K+qw5oQCJ8PP/WAOV1gWjM48Ju1oxNCCJGH9evXj48++ogaNWrc1t66dWsmTpxoSVB2794NGDOhXl5e2NjYMGfOnIdeTAbGzOT27dupUKGCJfnz9PQkPj4+y4u3ypYty6+//kqfPn04cODAPfs0adLEUj6wfv16PD09b5ttzoitra0lqRo9evR9+44cOdJSMwvG5/fdd9+RmpoKGCUfCQkJlC1bloiICJKTk4mNjWXt2rUZjhkXF4eXlxepqan3LX+4l7fffpupU6da/jHRokULFi1aZNnhIiYmhlOnTgFgb29vifNOH3zwAbGxsUyYMCHDZ/3999+WHSPi4uI4fvw4ZcqUua3P5MmTiYuLy3AnjIzUq1ePDRs2cPXqVUwmk+UnAZnZtGkTFSpUAKBVq1a31SyHh4cD4OrqettM8vHjx6lfvz6jR4/G09OTM2fOkJMkQX7Etl9zZXFyfdj0Ffz2OlzYC7FRkJoIvwTD4ZVwy79WhRBCiJu8vb0ZOnToXe0ffvghqamp+Pv74+fnx4cffgjAwIEDmT17Ng0aNODIkSO3zXhm1c0aZH9/f2rUqEGXLl1wd3fn1VdfpUaNGnTu3NlSCpAVlStXZt68eXTr1o3jx4/fdT0kJISwsDD8/f0ZMWIEs2fPvsco2dOuXTtu3eK1f//+VKtWjdq1a+Pn58drr72GyWSidOnSdO/eHX9/f3r16kWtWrUyHPOTTz6hfv36tGzZkipVqjxQPJ6enjz33HMkJycDRsnCmDFjaNWqFf7+/rRs2ZLz588DxgzxzXhuFRUVxaeffkpERAS1a9cmICCA6dOn3/WsnTt3EhgYiL+/Pw0bNqR///53/f6NHz+effv2WWbkp0yZkqX3UapUKf7zn/9Qv359nnnmGapVq2Ypw7jTzRpkf39/du/ebfkz+80331h+/6tVq2Z5docOHViyZAkBAQFs3LiR4cOHU6NGDfz8/GjSpAk1a9bMUoxZpR5mOvxxExgYqG+u1sxtaemaz/88xJqNm3jWO4lB/Qfg5OgIKQnGLPKVw1CqDrz0KxRwfyQxCSGEyNjBgwepWrWqtcMQIl+Ij4+nYMGCmEwmnnvuOfr168dzzz1nlVju9XdXKbVTax2YwS0WMoP8iNnaKP7Trir9O7dm4pnyjFx6yPiRmIMLvLwSWv8Xzu+FaU1h3WcQc8LaIQshhBBCZElISAgBAQH4+flRrlw5OnfubO2QHkpe2Af5ifRi/TJcuJ7EN2uP4lnQgdQ0TXCQD2UaDoTi1WHNKNjwBeyeC71/g6KVrB2yEEIIIcR9PegJh3mVzCBb0ZDmFanqVYip/5zgh80n6Th5EzEJKVD+aRiwHl77B25cg8l1Yetka4crhBBCCPFEkATZiuxsbZjYM4ChLXxZMjCIuCQTQxfs5p8j5m3gvPxhcChUagurP4Kzu6wbsBBCCCHEE0BKLKysYjFXhrU09vd7u2UlJv59lI1Hr+DubE85Txfeb1uVes99B989BYv7G7PKjgWtHLUQQgghxONLZpDzkEHNKrJ3VGveblmJZ6oW59L1ZF6YtpUaY3fwc+kPIOa4sT2cEEIIIYTINZIg5zEOdjYMaeHL+G41WTWsCW+1qETdch68u9ONSK92sHUSxGR+PrwQQojHh1KK3r17W16bTCaKFi1K+/btc/W5wcHBlCtXjoCAAKpUqcLHH3/80GPNmjWLwYMH58j9ISEhD7UY7PLly9jb2zN16tTb2n18fLhy5cpDx5YdISEhlCpVyvIZv/HGG7ed3ncvv/32223HMWc07p369etHsWLF8PPzu609JiaGli1b4uvrS8uWLbl69eoDv4/HjSTIeZiLox1Dn/FlRt9AAssW5s3LHUm3cyL2hy68O/Yrts7+D9uPXSIh+f7HgwohhMjfXFxc2L9/Pzdu3ABg9erVlCpV6pE8e9y4cZYT6mbPns3Jk7k3SZPZcdfZ9csvv9CgQQPmz5+fq895UMOGDSM8PJyIiAj27dvHhg0b7tv/fgnyuXPnaNu2Ld9++y0BAQF89dW/P3kODg7mzz//vOuesWPH0qJFC44ePUqLFi0YO3Zs9t7QY0AS5HxAKcW7baqwP6EQL14fjENcFF8khdDw5GQWzRxH0/HrOXvthrXDFEKIx9/KETDz2Zz9Wpm143zbtm3LihUrAJg/fz49e/a0XEtISKBfv37UrVuXWrVqsXTpUgAiIyNp3LgxtWvXpnbt2mzZsgUwjnBu2rQpXbt2pUqVKvTq1YvMDg67ebz0zdP4Ro8eTd26dfHz82PAgAGW+5s2bcp7771HvXr1qFSpEhs3brxrrBUrVtCwYUOuXLlCcHAwb7/9Ns2aNeO9994jJiaGzp074+/vT4MGDdi7d2+WPp+smD9/Pl9++SVRUVGcPXv2rutffPEF33zzDWAkrc2bNwdg7dq1vPTSSwC88cYbBAYGUr16dUaNGmW5futhGKtXr6ZLly6kpaURHByMn58fNWrUuC1ZvZeUlBSSkpIoXLgwYByn3KZNG+rUqUPjxo05dOgQW7ZsYdmyZZYTDu88jXDChAnUqVOHgQMHEhoaSps2bSzXmjRpgoeHx13PXbp0KX379gWgb9++/Pbbb/f/IJ8AkiDnE/XKebDizcY0a92FdQ1mYAroQ2oxfz5xXYJDSizPf7uFhaGnrR2mEEKIXNKjRw8WLFhAUlISe/fupX79+pZrn376Kc2bNyc0NJR169YxfPhwEhISKFasGKtXr2bXrl0sXLiQIUOGWO7ZvXs3EyZMICIighMnTrB58+Z7PvdmIubt7U2PHj0oVqwYAIMHDyY0NNQys/37779b7jGZTOzYsYMJEybcVZaxZMkSxo4dyx9//IGnpycAR44cYc2aNXz55ZeMGjWKWrVqsXfvXj777DP69OmTI5/fmTNnuHDhAvXq1aN79+4sXLjwrj5NmjSxJPRhYWHEx8eTmprKpk2baNy4MWB81mFhYezdu5cNGzawd+9emjdvzsGDB7l82diFaubMmbz88suEh4dz9uxZ9u/fz759+3j55ZfvGdtXX31FQEAAXl5eVKpUiYCAAMA4VnrixIns3LmT8ePHM3DgQIKCgujYsaNlZr9ChQq3jeXg4EB0dDSpqanY29tn6RTIixcv4uXlBYCXlxeXLl3K4qf6+JJdLPKRaiULUa1kIaAC0BHO7sJ+RitWlJnHK0lvMeLXfZQu7ExQRU9rhyqEEI+nttb70bO/vz+RkZHMnz+fdu3a3XZt1apVLFu2zFKXm5SUxOnTpylZsiSDBw8mPDwcW1tbjhw5YrmnXr16eHt7AxAQEEBkZCSNGjW667njxo2ja9euxMfH06JFC7Zs2UJQUBDr1q3jiy++IDExkZiYGKpXr06HDh0A6NKlCwB16tQhMjLSMta6desICwtj1apVFCpUyNLerVs3bG1tAdi0aROLFy8GoHnz5kRHRxMbG5vdj48FCxbQvXt3wPjHxiuvvMLbb799W586deqwc+dO4uLicHR0pHbt2oSFhbFx40bLzPLPP//MtGnTMJlMnD9/noiICPz9/enduzdz587l5ZdfZuvWrfz444/ExcVx4sQJ3nzzTZ599llatWp1z9iGDRvGO++8Q2pqKl27dmXBggW0b9+eLVu20K1bN0u/5OTkTN/n8OHDeffdd5k6dSrbtm3jww8/pGnTpg/5qT25JEHOz0rVhtaf4b5yOAsCy9P6Rnve+3Uvq4c9jZO9rbWjE0IIkcM6duzIO++8w/r164mOjra0a61ZvHgxlStXvq1/SEgIxYsXZ8+ePaSnp+Pk5GS55ujoaPm1ra1tpvW/BQsWpGnTpmzatInatWszcOBAwsLCKF26NCEhIZYSjFvHvnPc8uXLc+LECY4cOUJgYKCl/WbZxs33ciel1H1jA0hLS6NOnTqA8TmNHj36tuvz58/n4sWLzJs3DzBqdY8ePYqvr6+lj729PT4+PsycOZOgoCD8/f1Zt24dx48fp2rVqpw8eZLx48cTGhpK4cKFCQ4Otrzvl19+mQ4dOuDk5ES3bt2ws7OjcOHC7Nmzh7/++ovJkyfz888/88MPP2T4Huzt7WnTpg3//PMP7dq1w93dnfDw8Ezf+63c3NyYOnUqXl5etG7dmk6dOnH69Onbfu/vVLx4cc6fP4+Xlxfnz5+3/JTgSSYlFvld/QFQ91Xsw6YzrqkjZ2Ju8N7ivVy6npT5vUIIIfKVfv368dFHH1GjRo3b2lu3bs3EiRMtyeXu3bsBiI2NxcvLCxsbG+bMmUNaWtpDP9tkMrF9+3YqVKhgSQo9PT2Jj49n0aJFWRqjbNmy/Prrr/Tp04cDBw7cs0+TJk0sSez69evx9PS8bbY5I7a2tpbFhHcmx4cPHyYhIYGzZ88SGRlJZGQk77//PgsWLLjn88ePH0+TJk1o3LgxU6ZMISAgAKUU169fx8XFBTc3Ny5evMjKlSst95UsWZKSJUsyZswYgoODAbhy5Qrp6ek8//zzfPLJJ+zadf8Dv7TWbNmyhQoVKlCoUCHKlSvHL7/8Yrm2Z88eAFxdXYmLi7vnGAcPHrTsglGjRg1sbGxITU2973M7duzI7NmzAZg9ezadOnW6b/8ngSTIj4Om74O9M3XC3mV0jWj+2Heedxbl3KIGIYQQeYO3tzdDhw69q/3DDz8kNTUVf39//Pz8+PDDDwEYOHAgs2fPpkGDBhw5cuS2mdqsulmD7O/vT40aNejSpQvu7u68+uqr1KhRg86dO1O3bt0sj1e5cmXmzZtHt27d7lpgBsasd1hYGP7+/owYMcKSuGXH/Pnzb1tEB/D888/fczeLxo0bc/78eRo2bEjx4sVxcnKy1B/XrFmTWrVqUb16dfr168dTTz112729evWidOnSVKtWDYCzZ8/StGlTAgICCA4O5r///e8947tZg+zn54fJZGLgwIEAzJs3jxkzZlCzZk2qV69uWXzZo0cPxo0bR61ate76DDdv3kxQUBAzZ86kfv36jBw5EldX40Cynj170rBhQw4fPoy3tzczZswAYMSIEaxevRpfX19Wr17NiBFZWzj6OFOZrVp9EgQGBuqwsDBrh5E9u+fB2tGQeoOpAb8w9p8rrB7WhL8OXKRzrVKUci9g7QiFECJfOnjwYJYWOgkxePBgatWqxSuvvGLtUAgJCbnnXshPknv93VVK7dRaB2Zwi4XMID8uavWCvssgJZ4+MROxJZ3Ok7cw7q/DfLL8/puJCyGEECJ76tSpw969ey3bwVmbLMzLHlmk9zgpWhlafEiBNSGsq+zI0ISXsbez5c8DF9hz5ho1S7tbO0IhhBDisbRz505rh3AbSZCzRxLkx02jYZAcT+mN4/m1YQlSL0TwbsGGDJjjSIWiBWnk60lMfAo+ni70ql8mSyuDhRDiSae1lv9eCpGPZLeEWBLkx1HzD+Daadg6CXtgvN1W+qjPibpWhi/+PIyDnQ0ppnQ8XBxoV8PL2tEKIUSe5uTkRHR0NEWKFJEkWYh8QGtNdHT0fbe2y4wkyI8jpaDTJChYDLzrYrt0MPN8/kLXe5VLTlXxKOFDh4mbGPN7BM0qF6OAg+yZLIQQGfH29iYqKspySpoQIu9zcnKyHITzMGQXCx6TXSzuZ00IbDKf/25jD32XsyO9Mt2nbuXN5hX5v1aV73u7EEIIIcTjQHaxEP8KGgK1ekPXH6CQF/z2BvW8nekUUJKp/5zg/IULcOkg/NgZNoyzdrRCCCGEEFYlJRZPAmcPo+QCoIAHzOkM26cwomFbSh1YhvOU/kC8cf3kBmK9m7LbVJbqJd0o6uqY4bBCCCGEEI8jKbHgCSixuNO87nB8LShbSEvmiH1lVqrGhCd48pXT98SZbOiR/AGX7YrzauNyVPNyo2KxgpQo5ISjvQ3Xb6RSrNDDF74LIYQQQlhDVkssJEHmCUyQ4y/D5gmQegOeGgqFy3I9KZVu323FM+4AP/Ax9pj4rcgAVp51JDy9AjEUQtnY4epkR1JqOv+820xml4UQQgiRr0iC/ACeuAQ5Azf/LKhrp+GPd+DoKss1k20Bfir5Ph8drQjAoGYVGN66ilXiFEIIIYR4GJIgPwBJkO/BlAyhM6BIRbh8ECKWwbndMGQXg/6IYf3hS/z5VhNKezhbO1IhhBBCiCyRXSxE9tg5QsOBUKmVUYbRbRbodNizgPfbVUEpRYv/bWDQvF3WjlQIIYQQIkdJgiyyxr00lGsModPxPjCNiT0CqF/OgxX7ztN7xnbWH75k7QiFEEIIIXJEnkmQlVJDlVL7lVIHlFJvmdtClFJnlVLh5q92t/R/Xyl1TCl1WCnV+pb2Nua2Y0qpEdZ4L4+toCFgVwDWjKLZtcVM61kDz4IObDx6hSHzd3M5LtnaEQohhBBCZFueSJCVUn7Aq0A9oCbQXinla778ldY6wPz1h7l/NaAHUB1oA3yrlLJVStkCk4G2QDWgp7mvyAm+LWHoHijXBP56nwI/tuLn/oFM7xNIUmo6I5fsQ2rahRBCCJHf5YkEGagKbNNaJ2qtTcAG4Ln79O8ELNBaJ2utTwLHMJLresAxrfUJrXUKsMDcV+QUGxvouRBajYEL+yh/ehHPVCvO8NaVWRVxkZ7fbyPi3HVrRymEEEII8dDySoK8H2iilCqilHIG2gGlzdcGK6X2KqV+UEoVNreVAs7ccn+UuS2j9rsopQYopcKUUmGXL1/Oyffy+HNwhoaDwacxrB4F5/fySqNyDG3hy9GL8Qz6aRc3UtKsHaUQQgghxEPJEwmy1vog8DmwGvgT2AOYgO+ACkAAcB740nyLutcw92m/1zOnaa0DtdaBRYsWzd4beBIpBc9PByc3WD4UGxvFsJaVmNizFievJDB+1WFrRyiEEEII8VDyRIIMoLWeobWurbVuAsQAR7XWF7XWaVrrdOB7jBIKMGaGS99yuzdw7j7tIje4loBGb8G5XbB0MGyaQFDxNPo0LMsPm0/Sa/o2Gny2lqMX46wdqRBCCCFEluWZBFkpVcz8vQzQBZivlPK6pctzGKUYAMuAHkoYWJGAAAAgAElEQVQpR6VUOcAX2AGEAr5KqXJKKQeMhXzLHtV7eCL5dze+754Da0bBzDa8X8+W7nVKc/5aEnFJqYz+PYLo+GQSU0zWjVUIIYQQIgvsrB3ALRYrpYoAqcAgrfVVpdQcpVQARplEJPAagNb6gFLqZyACoxRjkNY6DUApNRj4C7AFftBaH3j0b+UJUqAw9JgPtg7gWBDmPk+BqQ34vNNk6PoSP2w6yejfI6gzZg0uDrY0r1qcc9du8GK9Mjxfx9va0QshhBBC3EWOmkaOms5R8ZdgUT84uwt6/4ouXZ+tx6OJOH+dRTujOHopnnKeLhy7FM/7bavw2tMVrB2xEEIIIZ4QWT1qWhJkJEHOcbFnYWYbuHYGOkyAmi+CnQM3UtKITkimRCEn3py/mzUHL7J62NP4eLpYO2IhhBBCPAGymiDnmRpk8RhxKwVvbIEKzWH5UPhfFYg+TgEHW7wLO2Nna8PHnarjYGvDG/N2EXU10doRCyGEEEJYSIIscoejK/T4CTp9C6Zk+GskpKVaLhdzdWJSr9pExSQybGG4nMAnhBBCiDxDEmSRe+ydoFYvaDIcjqyEcRVgcn0jWdaaZpWL8W6byoRGXmXd4UvWjlYIIYQQApAEWTwKTw01jqeu3sXYO3nrJAj/CYAX6pahvKcLHyzZT2xiaiYDCSGEEELkPkmQRe5TCiq3MRbsvbQEyj4FK/4PIpbhgImvXgjgUlwy/WaHEh2fbO1ohRBCCPGEkwRZPFo2NtBtNhQqCT/3hl/7U7O0OxN71mJv1DWeHreetxeGE3tDZpOFEEIIYR2SIItHr2BReH0jNBgIEUvh4gHa1vBi5dDGtK5egl93n2Xe9lPWjlIIIYQQTyhJkIV1OLjA0++Cg3H6HtunUdENvuxek7o+hfniz8P0nrGd57/bQlySzCYLIYQQ4tGRBFlYT4HC0GcpFC4HK4fDF+Vh3yK61DaOoD5w7jo7T11lzrZTxCebOBMj+yULIYQQIvfJSXrISXpWpzWc2Q6rP4IL+0jr/zd/R3vQ2NeT1+bsZE/UNbzcCnDownUa+xalTfUS9KxXGqUUu09fpXzRgrgVsLf2uxBCCCFEHidHTT8ASZDziOvnYUojcCkKr/4NDs4cuxRHr+nbuRyXzHO1vAk/c5XjlxMoXsiRQk72HL0UTxkPZz7uVJ1GFT05fjke32Ku2Nooy7CxiakkpppwsLXBztZGkmkhhBDiCSUJ8gOQBDkPObbWqEmu1Qs6TQbgSnwyUVdvEFDaHa01v+yMYtuJaKLjU6heshBLdp/lfGwSzg62JKakUb6oC7XLFCb2Riol3ZxYtucc8ckmUtM0dX0K88vrQVZ+k0IIIYSwhlxJkJVSzkBh88urWuvHoihUEuQ8Zu1o2PglVGoDbf4LHuXv2z3FlM4f+86z9tAlqnkV4u9DFzl+OQG3AvZcup5EQBl3yni4MH/HaQCWDnqKmuZkWyl137GFEEII8fjIsQRZKVUSeBfoBJS54/JpYCkwTmt99iFjtTpJkPOY9DTYOhk2fA5pqcZR1U3eMQ4cyYa4pFQafLaW6qXcsFWKnaev0rFmSUZ1qIark5RdCCGEEI+7HEmQlVJ+wDqM3S6WAweAGEBhzCRXAzqYuzfVWu/PZtxWIQlyHnX9HKx8Dw4ugx4/QZVnsz3kop1RvPPLHgo62tGqWnGWhJ+liIsjz9cpxdAWvjg72OVA4EIIIYTIi3IqQV4N2AMdtdbXM+hTCFgGpGitWz1kvFYlCXIelpYKUxrDjavw0mIo4ZftITcdvUKpwgUo5+nCnjPX+HrtUdYdvkTvBmUZ3Sn74wshhBAib8pqgpzZPsgNgc8ySo4BzNf+C8jKJ5HzbO3h+elGecXcLmBKzvaQjXw9KefpAkDN0u78EFyXvg19+HHrKfr8sIPrcjCJEEII8UTLLEG+AbhnYRx3ICn74QhxDyX8oPO3EH8RDvyWK48Y0bYKb7esxOZjVxg0bxdTNhwn6upjsQZVCCGEEA8oswR5KTBeKdUkow5KqcbAF0DuZC5CAJRrCkV8YcXbMK4i/Pl+jg7vZG/LkBa+DGnuy8ajVxi78hDNx29g+sYTOfocIYQQQuR9mdUgu2MszgsCzgH7gauABjyA6kApYBvwrNb6Wm4HnBukBjmfuLAfQr+HnbPA3hmGHwMHlxx9hNaa60km4pNNjFp6gDUHLzLz5bo0q1zM0ifi3HXKFnHGxVEW9AkhhBD5SY7ug6yU6oSxW0V1jMRYYexmcQBjgd4ynY9PHJEEOZ85uRFmt4dSdSDoTajSAWxzPllNSk2j06TNXLiexKgO1ahU3JXElDRemLaVwLKFmfNKfZzsbXP8uUIIIYTIHXKS3gOQBDmfSU+Hb2rCNePgDxq9Dc+MunffxBg4/rdxfHX5px/4UWdiEuk2ZSsXrv9bYu/h4kBMQgr1ynng6mhHFS9Xjl9KIKCMO682Ln/bMddCCCGEyDskQX4AkiDnQ4kxxoEiK4bByX+g+49w6A9oOAgKlzXKMEKnQ/RxSE0EWwd4fTMUrfTAj4pLSuVUdCIR569z7FI8HfxLsu9sLCN/20dhZyNZLlHIiQvXk+hZrwyfPecnJ/QJIYQQeVBOnqRXAHgN4yS9atxy1DQQgbGQb1p+PnZaEuR87MwOmNHy39fuZcHTF46tgZK1jK+qHeGXYHD1ghcXGgl0DrgSn4yHswNX4pMp6urIuL8O8+3643Sr402zKsVoWa049raZrYMVQgghxKOSUweFlAb+BnyAzdz7JL2ngFNAC6316WxHbgWSIOdjWsOBJaDTwdnDOHkvMRqChhj1yTbmGuET62FhHyhYDF7fCPYFciEUzYdL9zN3m/HXQCnwLlyAAY3L07uhj6XPrtNXqV7STeqXhRBCiEcspxLkxYAvxkl6kRn08cHY4u241vr5hwnW2iRBfkIcXwdzOkPlZ4165I3/gxYfQq2XcvQxiSkmth6PJvzMNbafiGFHZAy1y7jTuVYpzsQk8v3GkzQsX4QZwYFytLUQQgjxCOVUghwLvKS1Xp7JwzoCc7TWbg8caR4gCfITZPM38PcnkJYCNvZQoDD0+gVK1Ph3tjkHpZjS+WjpfkIjYzh+OQGAxr6ebDp2hbZ+JRjfraYlSU5P1+w7G4uLoy0Vi7nmeCxCCCHEky6nEuRrQO8sJsg/aq2zcupeniMJ8hMm/jIkxULCJZjZ1mhzKw1dvoeyDXPlkaa0dKZvOknl4q40rVyU7zee4LM/DuFoZ0PTykVp71+SedtPse1EDDYKnvUvSVCFInSsWRJTuqago53sjiGEEEJkU04lyIuASkAnrfXJDPr4YCzUO6K17vZQ0VqZJMhPsDOhEHMc/hlnbBtXtz+0/CRX9lW+046TMfyx7zy/7z3HlfgUHOxsGNmuKkcuxrE64iKX4pIpYG9LkimNysVdsbNV/KddVYIqeOZ6bEIIIcTjKKcSZG9gHcYivW3c+yS9BkAk0FxrHZXdwK1BEmRBQjSsGgl75kObsdDgjUf26KTUNI5ejKeEmxNFXR2Bm4v5rvHrrigc7WxZc/Ai0fHJFCvkxIohjaR2WQghhHgIOb3N2wBuP0kPjET55kl638s2byLf0xrmPg9RofD6phzbDi6nrD98ieCZobg62vHjK/XwLuxMYoqJS3HJBJYtLHsvCyGEEJmQg0IegCTIwiLmJExtAp6V4KVFkGaCgkWtHZVFWGQMQxeEczk+mRRTuqW9Zml3KhcvyKW4ZHrULU3r6iUkYRZCCCHukO8SZKXUUOBVjD2Wv9daT1BKeQALMUo8IoHuWuuryvg//9dAOyARCNZa7zKP0xf4wDzsGK317MyeLQmyuM2B3+CXvmDnBI6uMDjU2O0ij9hxMoaRS/bxrL8XhZ0dcLCzYfxfh7melEoxVyfOXrtB7TLueLg40MbPi651vK0dshBCCJEn5GiCrJTqzL8n6d0ssYjBfJKe1vq3bMSKUsoPWADUA1KAP4E3MBLmGK31WKXUCKCw1vo9pVQ74E2MBLk+8LXWur45oQ4DAjHqpHcCdbTWV+/3fEmQxV1Wf2QkyrFnoP7r0Oa/1o7ovmJvpBKXlEqJQk4s2hnFN2uPEpdkIi7ZxJSXatPGz8vaIQohhBBWl1OL9AoDy4Eg4DT3PkmvLMYCvmczS0Tv85xuQGutdX/z6w+BZOAVoKnW+rxSygtYr7WurJSaav71fHP/w0DTm19a69fM7bf1y4gkyOKetIZfX4Wjq+HdE7myT3JuSjal0WnSZpJN6awe1gQ7OfZaCCHEEy6rCXJm/8f8EigDPK219tFaP6u17q21fsn863JAE6AkMD4b8e4HmiiliiilnDFmhksDxbXW5wHM34uZ+5cCztxyf5S5LaP2uyilBiilwpRSYZcvX85G6OKxpRRUagNJ1+DcbmtH88Ac7WwZ1rISJ68kMPr3CK4lplg7JCGEECJfyCxB7gi8o7XemFEHrfUm4D2g88MGobU+CHwOrMYor9gDmO5zy71WH+n7tN/rmdO01oFa68CiRfPOIiyRx1RoDig4ttbakTyUVtWKExzkw49bT1Hvs7WsOnDB2iEJIYQQeV5mCbIjxnZumbkGOGQnEK31DK11ba11E4wyjqPARXNpBebvl8zdozBmmG/yBs7dp12Ih+PsAaXqwH7zjhb5jFKKkI7V+f3NRlT1KsTg+bsZ/9dhTGnpmd8shBBCPKEyS5C3AiOVUq4ZdTBfex/Ykp1AlFLFzN/LAF2A+Rh7LPc1d+mLcWIf5vY+ytAAiDWXYPwFtFJKFTbXT7cytwnx8Bq9BVeOwLZvjbrkfMivlBszg+vyTNViTFp3jPk7TluumdLSOX45nrikVCtGKIQQQuQdmR3H9RawHjillFrBvU/SexZIA5plM5bFSqkiQCowyLyd21jgZ6XUKxiLBG8eZf0HRp3yMYxt3l4G0FrHKKU+AULN/UZrrWOyGZd40lVpb5RarP4Qzu2CjpPAsaC1o3pgHi4OTH6xNj2/38b4VUfw8XThfGwSk9cd41R0Ip4FHZnwQgBPVSzC4YtxeLkVwK2AvbXDFkIIIR65rJyk5wW8i3GSXjn+rfPVGHsTLwPGaa3zbSmD7GIhMpVmgi1fw99jwK8rPP+9tSN6aCevJPDS9O2cvXYDgMrFXendsCzfbzzBqehEGvt6svnYFYoXcmJ630Cql3SzcsRCCCFEzsiVg0KUUk4Y27sp4KrW+sbDh5h3SIIssmztJ7BxPLyyGkrXg7iLULCYseNFPnI1IYXNx6/gXdiZmt5uKKW4kZLGt+uPMfHvY5T2KEBamiYlTfPboCA2Hb3C1H9O8GO/epT2cLZ2+EIIIcRDyXcn6VmTJMgiy5Lj4ZtaULQyNHgDFrwI5ZtC15nGgr7HwB/7zlPVqxBp6el0nryFwi72nIkx/i0cHORDSMfqVo5QCCGEeDg5tQ9yVh/WRClVOSfGEiJPcywIjYZB5Eb47Q1wKwOntsCilyH+EqQkQnqataPMlnY1vCjn6ULFYq6MfLYqZ2Ju0KpacToFlGT+jtN8vPyA7IIhhBDisZbZIr2sWg9opdRK4BOt9fYcGleIvCfwZWOx3uXD0P4ruHwIlg6G8b7G9YIloPtsKNPAunHmgB51S+NX0o2qXq5ciU8hKTWNmZsjiU1MxdvDmfrlPAiqUASVz0pMhBBCiPvJkRILpVRfwAVoBDTSWpfJ9qCPkJRYiGy7dBCO/GlsA7d7DiRGw1v7wOnxWuCmtWbwT7tZse+8pa12GXfeb1eVuj6PR4mJEEKIx5fUID8ASZBFjjq3G6Y1heYfQpN3rB1NjktKTePstRuUci/A4l1RTPr7GBeuJzGyXVX6Ny5v7fCEEEKIDD3SGmQhxC1K1gLf1rD5a4iNsnY0Oc7J3pYKRQviZG9Lr/plWfP207SpXoJP/zjI73vPkZSaxqnoBDpN3syinVHsjbpm7ZCFEEKIB5KlGWTzKXdtgWoYB4SAcRx0BLBSa30po3vzA5lBFjku5iR8FwQeFaDzZPCqae2IclVSaho9v9/G7tNGMlzQ0Y745H+P5p7yUh3a+JWwVnhCCCEEkEMlFkopG+BTYBjggHFq3VWMfZDdAWcgBZgAvK/zab2GJMgiVxxeCcveNOqR6w2AZiPBqZC1o8o1SalpzNh0kqsJKfwWfo53W1fG0d6GiX8fw5SWztJBjXBzlpP5hBBCWE9OJcgjgQ+Az4A5WuvIO66XBXoDI4ExWutPsxO0tUiCLHLNjavG4SJhP0DB4tDsPxDQC2yzsIFMchzMfR7KBsEzIbkdaa7ZdPQKL8/agaOdLTVKuTG1Tx0KOUmiLIQQ4tHLqQQ5EpigtZ6QycOGAUO11j4PGGeeIAmyyHVnd8Ifw43vNV+Ezt9mfPpeehocWgHbvoXTW422hoONGWiH/HmKXfiZayzYcZpfdkbxTNViTHqxNva2sgRCCCHEo5XVBDmzaaziwO4sPG+Xua8Q4l5K1YH+a2H9f2HD5xB7BhJjICkW+iwFz4r/9l06CPbMN2ac238F58Jh6yQ4ucGYSY49CzV7gJ2jtd7NAwso7U5AaXcqFivImBUHeWPuTqb2DsTWRvZPFkIIkfdkliAfBHoAGzLp1xM4lCMRCfG4UgqeHgEOLrBjupEUx56BlcPhxV+MsoukWNi/GGr1hg5fg42tcW+V9vDrq0bJBYApGeoPuH18Uwqkm/L0LHP/xuWxtVF8vDyCb9cd480WvtYOSQghhLhLZgnyaGCR+RjpucABjEV6GmM3i+pAL+BpoGsuxinE48HGBp4aanwBbJ9mJMjTmxtlFHvmQ1oK1O77b3IMUKmVcfDIyQ2wcgRsnwKuJeDiAajTF5QN/NgJUm/Aq3+Di6d13l8WBAf5sPv0Nb5cfQSPgg6ka4g4F8v2EzH0a1SOlxqUtXaIQgghnnCZbvOmlGoN/BcIwEiMb7sM7AH+o7VemSsRPgJSgyysRms4sAT+HAHxF40297IwdE/GNcoRy+Dn3v++dnQD1+JG6YVOg0IlocM3UK5x7sf/kJJNabw+ZyfrDl8GoJCTHa5O9ly4nsTcV+rTsEIRK0cohBDicZTjJ+kppbwxZow9MBLjGOCA1vpMdgLNCyRBFlaXdB2uHAVHV2MrONdM9gy+sA/iLoB7GfipO1w7Ay8tMmaSlw81SjCG7IaUBIg+Bic2QJ1gI5HOI5JS03hv8V5KFHJiRNsqxCeb6Dx5M9cSU/lt0FPY2ihKuhewdphCCCEeI3LU9AOQBFnkawnREHcOStQwXkduglnPQukGxq4Z6alGe5kg6DEPnD0yHsvKjl2Kp8u3m0kypZNiSic4yIeP2lfDRhbzCSGEyAGP9KhppZSTUqpMTowlhHhALkX+TY4BfBrBU2/B9bPg9zz0+MnYDeP0FhhfCZa/ZdQq50EVixXkp1cbUKdMYdrVKMGsLZEs3vX4HdcthBAib8uRGWSl1PPAz1pr20w750EygyyeCOf3wM7ZxqElVZ6FjhPh+jko4WftyO4pPV3TdcoWIqMTaetXglPRifSqX4a2NbysHZoQQoh86pHOIAsh8gGvmtD+f9D2Czj0O/yvGkx5Cla8Y9Qz5zE2NorPn/fHrYA9P+04zckrCbwxbxeLdsqMshBCiNx1323elFIfZXGcajkQixDiUag/wDgCO3wulG0EYTNg38/GzhfVOmW8e4YV+BZ3ZdWwJsQnmXB2tOWVWWG8t3gvLg62NKtSDDsbhZ2cyCeEECKHZXbUdDrG1m5Z+T+mlhILIfKhK8fgl2C4uA+qdoBuPxr7NedBCckmek3fTviZazja2dCyWnG+7F4TOxsbOZVPCCFEpnJkFwul1DlgCTAkk3GeB+ZLgixEPpWWCpu+gnWfQvMPoFpncPOGy4fh7zHQ+lMoWtnaUQJwIyWNb/4+yq5TV9l+MgYHWxu83J3o29CHmqXdSTGlU6+chyTMQggh7pLVBDmzk/TCgNpa67RMHnbf60KIPM7WHpoMhytH4O9PjaS4SEVjEV9qIuh06P2rtaMEoICDLe+1qUKyKY3eM3bgWdCBi9eTGf17hKVPQGl3hreuzFMV8+6JgkIIIfKuzBLk9UD/LIwTCfyY3WCEEFakFHT4GhJjwLkIHPkTyj0NxaoYs8uH/oAq7Yy+KYlgX8Cq9cqOdrb8/FpDy+tjl+I4HZPI5bhkvlx1hF7Tt/NMVaMEw62AvdXiFEIIkf/IQSFIiYUQ95SeBja2YEqB75tBwmV4bips/BIiN4JrSajcFjzKQ+DL4OBi7Ygtkk1pzNocyfhVh/Et5srSwU9hL4v5hBDiiScn6T0ASZCFyMSFfTDzWUiOBSc3qPsqnN4GF/ZC8nUoXA5ajoaKLfJUorx8zznenL+bD56tyrlrSSSmmHi7ZSWKFXKydmhCCCGsQBLkByAJshBZcPkw7F9sJMcFi/7bHrkJFr0C8RegaFVjVjmgF3hWtF6sZlprOkzaxP6z11EK7G1s8PYowIo3G1PAIV+uKRZCCJENOZogK6UqAO5a6523tLUEvgGKAcuAwVrrhIcP2XokQRYim1IS4fha+G2QMcvsWQle/RscXa0dGUcvxvHP0SsEVShCTEIKvaZvp6CjHa80KsewlpWsHZ4QQohHKKcT5DXAca31a+bXRYDjwCEgFAgGvtdav52doK1FEmQhcogp2ahPntcNCnnDS4uhsA/YOVg7Moul4WdZtDOKjUevMPnF2jzrL0dXCyHEkyKnj5quCay+5XUXIAVoprV+E3gbeO6BoxRCPF7sHKHiMxD8B6QmwMy2MLY0nNpq7cgsOgWU4ofguvh7u/Hx8gNsOxHNjRTZqVIIIcS/MjsoZJT5lx8BM4HTGKfqtQMKA3PN14sBA4Ax5tfrtdb/5EbAuUFmkIXIBQd/h5/7gL0zuHhC8erGvso9F4BrcWtHx+7TV3nu2y0AODvYUqKQE6M6VufpSkUzuVMIIUR+lVMn6T1t/uVfwHtAuPn1POAX4Dfz61LAdKANRgIdqbU+9XChP3qSIAuRS1ISIWoHLHkDdBokxxklF11nQuGyEDbT2Ge5QnOrhLf79FUuxyWz8egVtp2I5uSVBF6sX4b/tKuKk70s4hNCiMdNTtcg7wGOAP2A5sCvQIDWep/5egtgitbaNxsBD8M4lEQD+4CXgSnA00CsuVuw1jpcKaWArzFmshPN7bvM4/QFPjD3H6O1np3ZsyVBFuIROb4OFr8CaSYoVBIuHwQnd3hzF7gUsWpo15NSGbvyEPN3nKZRRU+m9Q6UnS6EEOIxk9M1yJ8AnYFrGMnxkpvJsVlXYOMDR2mmlCoFDAECtdZ+gC3Qw3x5uNY6wPx1cwa7LeBr/hoAfGcexwMYBdQH6gGjlFKFHzYuIUQOq9AMBqw3TuG7dgrafgEp8fBDK9gyETZ/DdHHrRJaISd7PnuuBl8878+mY1fo/2MoKaZ0q8QihBDCujI7ahoArfUipdQxoClwHlh0R5eTGGUY2Y2lgFIqFXAGzt2nbyfgR21Mf29TSrkrpbzM8a3WWscAKKVWY5R9zM9mbEKInOJexkiSTUngUQ7cShun860y/+Dnny/B7zkoUQNq9wXbR3tMdLfA0gAMX7SXj5cfYExnP5QVj9QWQgjx6GUpQQYwz96GZ3Dti+wEobU+q5Qaj7EI8AawSmu9Sin1IvCpUuojYC0wQmudjFHzfOaWIaLMbRm130UpNQBj9pkyZcpkJ3whxIMqdMvWalXaGV8xJyD1Bqz9BA4sgZ2zjNfVOhlJ9SPULbA0xy7HM3XDCextbRjRtorUJAshxBMkqyUWucpcBtEJKAeUBFyUUi8B7wNVgLqAB8ZCQTAWAt5J36f97katp2mtA7XWgUWLyqp1IazOo7yx08WLC+C9U+DT2JhVnuAPZ3c98nDea12F4CAfZm2JxP/jVdQY9Rchyw488jiEEEI8enkiQQaeAU5qrf+fvfsOj7LM+jj+vSe9kwRCSUJJQJBeQhUEBMUKIhZQFxTsYFvdfV13raura1nXigqCBZRFREQFBGkC0nsvoQZCKCGBENJmzvvHM2BAhAQmeWaS87muuZy555nJjz+WPbk5z7kPikghVp9zJxFJF0s+1pi5du7r04DEYp9PwGrJ+KN1pZQvMQaufgWSukNoLHw3DL4bCtOehhLcWOwJDofh+d5NGHdfB+7uVJeUutF8+utOpq1LL5efr5RSyj7eUiDvBjoYY0LdEyp6ABvdfcW4124E1rmvnwwMNJYOQLaIpGP1QV9ljIl270pfxcX3Riul7FCjGQycBNe/BcfSYc14WPQ+bJ9drjE6JMXyt2sv5eOBKTRPiOLP41ezNi37/B9USinls7yiQBaRxVg3/q3AGvHmAD4Gxhpj1rrXqvLbQSRTgO3ANmAE8JD7ezKxJm4sdT9ePHnDnlLKRzXuDf+3A57aA1G1Yer/wfHD5R4jwM/ByIEpRIcGcvenSxizaBf7sk6Uew6llFJlr0RzkCs6nYOslI/YMQ/G3gxVG8DAyRAaU+4Rth3I4fYRizhwLB9/h+Hlvk25ra3e6KuUUr7AoweFVHRaICvlQ7b9DF8NsA4auWkkJLYt9whFThe7M3N5bvJ65m09RJXQAP7UoQ57s07Qr3UCl9WvyrYDOazak0X7ejEkxoSWe0allFK/VyYFsjEmCbgVqA0En/G2iMiQUqX0ElogK+Vjdv0K394PRQXQ41mIuxTiW5d7jLxCJ18t2c3EFXtZuzebQH8HBUUuhnZPZtLKfezNOkFEsD8N4sLZeTiXIZ3rMbR7/XLPqZRSyuLxAtkY0wf4Gqs/+ACQf8YlIiJJpQ3qDbRAVsoH7VsFI64AcVqvr38LUgbbEiW3oIjF2zPpkBTLM9+tY8LyNADeuq0F78zcRkGRizqxofyaepiP/tSGXk1q2JJTKaUqu7IokNdinaJ3h4gcvMh8XkULZKV81LUCP7AAACAASURBVJ4lYPxgxjNwcDM8ugqCIuxOxezNBziWV0TvFrUoKHLh5zAUuVz0G/4r+7PzmfnnrkSFlu8JgUoppUpeIJdmikUS8EZFK46VUj4ssR0ktIErX4TcQ9aEiyUjrN1lG3VvGEfvFrUACPR34OcwBPn78epNzTmSW8ArUzfamk8ppdS5laZA3gTEllUQpZS6YAkp0HEYrBoLU56Ej7taN/N5mabxUdzTpR7jlu5h3lbda1BKKW9VmgL5r8DT7hv1lFLKu/R4Dnq/Bw8sgNj6MOUvsH8d7Ftpd7LTPNbjEhrEhfPouFWs36cHjiillDc6Zw+yMeaXM5bqY+0ibwXOPIBDRKSrZ+OVD+1BVqqC2TkfvrgJnO57iR9dDdF1bY1UXOrBHPp/vIiDx/KpGRXMsCvq079tbfwcxu5oSilVoXnkJj1jzBygxHPgRKR7Sa/1JlogK1UB7VkKiz+EdROg7T1w3Zt2JzpNVm4BXy3Zw+xNB1iyM5PEmBAaVo/g3i5JtE/SbjallCoLelBIKWiBrFQF9v2jsPwz6PUyxKdYp+9VbWB3qlNEhMmr9/H96nRW7cniUE4+nZJjeeqaRjRPqGJ3PKWUqlDKYszbQOBHETl8lvdigOtF5PNSJ/UCWiArVYEVnoCv74It06zXgeFw9avQ7BYIOPO8I3udKHDy5ZLdfDg3leP5RXwyqC3JcWEEB/gRGaxj4ZRS6mKVRYHsBDqKyJKzvNcGWCIifqVO6gW0QFaqghOBLT9Zo+AWfwT710BCO+j0MCR1heAouxOe5sDRPPp+8Ct7s06cWmtfL4a2dWPo3y6RhGg9uloppS5EWRTILqDDHxTIlwMzRCSo1Em9gBbISlUiLpfVl/zt/SAuaHIT3DLa7lS/k5VbwPQNGeQVOjlyvJDxy/aQnn2CmLBAxt/fkaRq4XZHVEopn+Opm/RaAq3dL0cCLwM7zrgsBOgPRItI0wuLay8tkJWqhA5tg2WjYNH7kDIEWg+EWi3tTnVO2w4c46YPfqVRjUgGdqrDlY2rE+Tvk/9wp5RStvBUgfwc8Jz7pQB/NIPoMDBERCaXNqg30AJZqUqqMA9+eBzWfWONhLv6VejwoN2pzumLRbt4ZtI6AFrXrsLIQW2JCQu0OZVSSvkGTxXIUUAVrMJ4O3ATcObU/XwgQ3x4HIYWyEpVcieOwHfDYNMP0KAX3DwKgryzhUFEWLQ9k31ZJ3j627XUiQ3lmwc7EaE38Sml1HmVtEA+50l6IpItIrtEZCdQD5jifl38sd+Xi2OllCIk2iqKr/gHbJ1u7Sof2Gh3qrMyxtAxOZZ+bRL4ZFBbth3I4dnv1uNy6V/DSinlKSU+atpdDBeUZRillLKNfxBc/he47BFYOx4+6AhLRtid6pw6N6jKIz0a8O3Kvdz3xTKcWiQrpZRH+J/rzeKj3dxTLM71t6+IyDm/TymlvF7PF6D5bTDrJZjyJBTlQ7v7wN87+3wf7dGAsEB/Xp6ykbdmbKF2bChfLt5N0/hIXuzdFIceX62UUqVWkpv0RojIPmPM85zn2GkRecGz8cqH9iArpX7HWQj/u9M6YKRWK+gwFMLjrLnJXkZEeGjsCqau2w9Azahg0rPz8HcY+rSM55WbmhHoX+J/MFRKqQpLj5ouBS2QlVJn5XLBmnEw6eRkCwPX/wdSBtsa62yKnC6W7zqCn8PQpk4045buYdnOI3yzIo1b2iTw+i0t7I6olFK2K2mBXOKWCGNMIxHZdHGxlFLKhzgc0PJ2OJEFhcdh10KY9jdwFkGTvhBeze6Ep/j7OWifFHvq9YB2tRnQrja1qgTz7qxttEiswp0d6tiYUCmlfEdpT9LLAGYDs4DZIpJahtnKje4gK6VK5Gg6fNAB8rKgZgsYPB0Cgu1OdU5Ol3DPZ0uZs+Ug1SOCee6GxlzTrKbdsZRSyhYeGfN2hl7AaKxxbx8AW4wxu40xnxpjBhljal9gVqWU8g2RNeGRldDvE0hfDZ/dAFl77E51Tn4Ow/t3tObh7vWpGhHI0C9XMG7JbrtjKaWUV7ugHmRjTDhwOXAF0ANojg9PsdAdZKVUqa3/Fr57GIyBxHbW9IsaTe1OdU65BUU8MGYFv2w5SMPqETgchgkPdCQsyJ9Ve7I4dCyfKxrF6eQLpVSFVaY36RljGgLdsYrjbkAMsFZEWpb6y7yAFshKqQtyOBXm/htSZ0NBDtwzE6o3tjvVORUUufj4l1TGLNrN/qN5xIQFcnObBMYu2sXxAid9Wtbi7f6t7I6plFJlwuMFsjFmMNaOcXegJrAZqxd5FjBHRA5feFx7aYGslLooR9Phw8vALxDiGsNtX0BgmN2pzmvIp0uZuenAqdcdkmJYtD2T0Xe1pXujOBuTKaVU2SiLAtkF5ALDgf+ISPrFRfQeWiArpS7a5qkw4zk4tBk6PARXv2J3ovPKyi0g9WAOS3Yc4WheIY/1bMD178xn/9E8Xr2pOeHB/jhdLr5elsbQ7vVpGh9ld2SllLooZVEgv4m1e9wcyALm4t5BFpGNF5HVdlogK6U85scnYOknMGS61ZvsY9KO5HL7iMXszsw9bT0y2J9nb2hCv9bxGKM9ykop31RmPcjGmGisQrkbVsvFpcABrEL5jtJHtZ8WyEopj8k/Bu93AOOAfiOgdgdrPXW21YJRp5N1Y58XK3QfOgKQfaKQOrGhPPXNWlbtyaJ/20SevaExoYE+eU+2UqqSK/OT9IwxdbFu0uvv/q+IiN8FfZnNtEBWSnnU3uUwfhBk74HoepDQFtZ+DQh0eQJ6PGt3wlJzuYQ3Z2zm/dmp1I0N5YdHuhAepEWyUsq3eHwOsjGmljHmTmPMJ8aYHUAq8CFQBXgduO6C0yqlVEUS3wYeWgi9/gXVGsHa8RBdBxrfCL++C5nb7U5Yag6H4S+9GjH6rrbsPJzLqPk77I6klFJlpjS//qcBAqwFJmH1H88VkaNlEUwppXxaUAR0HGo9dv0KkfFWi0XqLBg/EO76EYJ976a37o3iuLJxdT6Ys43kauFc11xP5VNKVTylOUnvFqCaiLQUkcdF5HtPFsfGmMeNMeuNMeuMMV8ZY4KNMfWMMYuNMVuNMf8zxgS6rw1yv97mfr9use/5m3t9szGml6fyKaXUBavTydpBjqwJN4+CjA3wbgpMfQpyM+1OV2ov921K45qRDP1yBc9+t46N6Uc5UeC0O5ZSSnlMiQtkEflGRMrkb3JjTDzwCJAiIk0BP6ze5n8Db4lIA+AIMMT9kSHAERGpD7zlvg5jTGP355oAVwMfGGN8si9aKVVBNbgS7vkZ4lvD0pEwqhcc960x8nERwYy7ryN3darLmEW7uObtedz96RK7YymllMeUZge5rPkDIcYYfyAUSMeakjHB/f5nwI3u533cr3G/38NYc4f6AONEJF9EdgDbAN+bs6SUqtjiW8Pt/4OBkyBzB/zwKFzgDdN2CfR38HzvJsz4c1fu6lSXRdszWbzdtwp9pZT6I15RIIvIXuANYDdWYZwNLAeyRKTIfVkaEO9+Hg/scX+2yH19bPH1s3xGKaW8S93O0OMZ2Pg9rBoLBcfhWIbdqUoluVo4T13TiKrhQbwxfTMXOhlJKaW8iVcUyO7Zyn2AekAtIAy45iyXnvyb92xDROUc62f7mfcZY5YZY5YdPHiw9KGVUsoTOg6DOp3hhz/DO63hvbaQvdfuVKUSHODHE1ddwtKdR5i4wreyK6XU2XhFgQz0BHaIyEERKQQmAp2AKu6WC4AEYJ/7eRqQCOB+PwrILL5+ls+cRkQ+FpEUEUmpVq2ap/88SilVMg4/uGU0NL0JYpPBVQjDO8KEwVCYZ3e6Ers1JZE2daL527drmbXJt3bBlVLqTN5SIO8GOhhjQt29xD2ADcBs4Gb3NYOA79zPJ7tf435/llj/rjcZ6O+eclEPaADonSNKKe8WHgd9P4S7p8CNH0DdLrDuG5g8zO5kJebnMIwcmEJytXAGf7qMTxfonGSllO8q8Ul6xpg+QIyIjHa/rgOMA5oCPwF3iUjOBQcx5gXgNqAIWAncg9U/PA6Ica/dKSL5xphg4AugFdbOcX8R2e7+nr8Dg93f85iITD3fz9aT9JRSXmfWy/DLa3D3VGtMnI/IK3TyyFcrmbExg0Ed63JFoziC/B0cOJZP5/pViQ4LtDuiUqoS8/hR08aYpcDXIvKa+/U3WBMixgN/Aj4XkScvPLJ9tEBWSnmdglyrH9lVCFe+CA5/SOoOYbF2JzuvvEIn/5i0ju9W7aXQ+dv/x0QG+/PT45dTMyrExnRKqcqsLArkTOB2EZlmjAnB2rkdKCJfG2PuAf4mIskXldomWiArpbzSgU0wph8cTbNeh9eAgd9BXCN7c5XQ0bxC1u89Sn6RkyB/PwaNWsINLWrxxi3NMcYgInw4dzvhQX7c2aEOVoedUkqVnZIWyKU5ajoYOOF+3sn92enu15uxpk8opZTylLhG8MhKOLwNcg9bR1RP/atVJPtAMRkZHEDH5N92vO/uXJeP5m5n/b5sYsMDKSwSluy0zp96/vsNXFozgvsvTybrRCG9m9ciKjTAruhKqUquNAXyTqAzMBdrJNtyEcl2vxeHNYtYKaWUJ/kHQvXG1vOuf4VpT8HmqdDoWntzXYC/9mpEragQft6YwYkCJycKnTzQNZn6ceFsPXCMmRsP8PBXKwGYsSGDJ668hN2ZuVzXrCYOh/f/QqCUqjhK02LxKNZhHquBlsCDIjLC/d4bQGsRuaKsgpYlbbFQSvmEogL4uBvkHoIHFkB4xRpRmVfoZNyS3Rw+XsC7s7adWr+9fW0C/RykHsxh5KAUgvz9bEyplPJlHm+xEJG3jTGHgA7AOyLyebG3I4BPS51SKaVUyfkHWuPgPrkKPu8NNw6HmHqw8QfYtQCu/691jY8KDvDjrsvqAdDz0urszsxlyY5Mvli069Q1o+bv5MFup9/usiczl2oRQQQHaOGslPKM0uwg1wbS3Qd5nPleAFBTRHZ7OF+50B1kpZRP2T7H6kfOywbjABFAoO290OtlwPh0oVyciLDtQA7GwKtTNzFr0wGeub4xd7sL6W0Hcrj27XnUjwvn1pQEejauTkJ0qM2plVLeqiymWDiBjiLyu4M3jDFtgCUi4pO/vmuBrJTyOccPwaYfIXM75GRAUAQs+RgCQqFqA0gZDPPfgtBYCK5irV32KETWgiO7YOt06xqH7/y1fSyvkMf/t5pZmzK4NSWR6pHBLEw9zPp92TiM4Vh+EcZA7ZhQXunbjE71q9odWSnlZcqiQHYBHf6gQO4AzBMRn7zlWAtkpZTPE4GZL8L+tbDtZ0CgVmsQl/U4uBkCQiAoEory4PgB6PSwtescXcfu9CV2NK+QK/8zlwPH8hEBf4fhX32bcWOrePZn5zFp1V6+W7WXHYeO06ZONO8MaKVzl5VSp3ikQDbGVME6xQ5gG9AP6ya94kKAYcD1IpJ4YXHtpQWyUqpCWT/J+m/jPr+Ng8tYD3NesQrpQ1shKgFSZwIG6naGS3tDlURrR7pGM0hoBwHB1mePH7J2or1ktNyuw8fJL3IBEBro97uWiqzcAkYt2Mno+TuoHhXMlEe6EOjvOO0ap0uYsjadlolVSIzRlgylKgtPFcjPAc8B59tmNsBzIvLPUqX0ElogK6UqHZcT9iyGbTOtVo2DG09/PzIBejwDYVVhzM0Q3xpuGwuRNe3JewGmrdvPA2OW897trbi+uTWq3+kSvlqymzGLdrFp/zHiIoJ48qqG9G5ZS2/yU6oS8FSB3AJrpJsBRgEvAalnXJYPbBCRNRce115aICulKjWXC2a/BEX50HEY7F0Ov7wG6autmwAj462DShLbwZ3fgsNx/u/0Ai6XcPnrs4mLCOK/t7UiONDBs5PWM239fprGR3Jjy3hGL9jJ3qwTXN+8Jo/2aED9uHA90U+pCsyjPcjGGD/gYWCsiBz0QD6vogWyUkqdweWCdRNgyQi48gXYvw6m/gXungZ1OtqdrsTGLt7F379dB0BYoB/HC5w80qMBj/dsgDEGl0t4b/Y2/jNjCwBXNa7O5ZdU49pmNYkJC2RrxjE27T9Gu3oxxIQFEuDnG78cKKXOztMFsgNrp/g6EZl+vut9jRbISil1HieOwGtJcPlfofvf7E5TKst2ZrJqTxbD56Ti72eY+5fup7VTuFzCj2vT2ZJxjI/mbqfA6SIs0I/Hr7yED+emciinAID4KiG83b8lKXVj/uhHKaW8XFlMsdgOPCEi315sOG+jBbJSSpXAiCvA4Q9DfHOfZH92HoVO1zlvynO6hK0HjvH0xLWs2J1FVEgAL/ZpwpHj1o1/+UVOZj3RjbCgEp+zpZTyIh4/SQ/4CHjMGPOjiBRceDSllFI+KakbzP8v5GZCqO/totaICj7vNX4OQ6MakYy/vyPbDx2nemQwUSHWBNNmCVH0G76QPu8voHHNSB7omkzjWpFlHVspZYPSFMgRQDKw3RgzDUjn9OkWIiLPeTKcUkopL9L0Zpj3Jiz6AK74xx9f53LB6i+tgjoqobzSeZS/n4NLqkecttamTgz/vLEp09fvZ+6Wg/yaepiHuiXTND6K71fv467L6pJcLdymxEopTyrtQSHnInqSnlJKVXDjB8GGSVC9KQz8zhoDd6YtP8GXt1rPH10N0XXLNWJ52HbgGHeMXEzG0fxTa6GBfnwxpD17s06w89BxOjeoSkJ0CFEhAfg7HGxMP0pUSAAJ0SE6KUMpm3i8B7ki0wJZKaVKKDfTmmwx702ITYYuT0Czm397Pz8Hxt0OO+aCf7C1i3z7/+xKW6ZEhNSDx5m4Io0el8bx8Jcr2Zed97vrakYFE+DnYHdmLgDVIoKoVzWMl25semqXeuS87UxYnsZng9sRGuhHoVOICQss1z+PUpWBFsiloAWyUkqV0sYf4OfnrJP3er4AO+dDXrY1O7noBPR4zrqhb8YzcOOH0HKA3YnL3Lq92Xy+cCfXNa9F8/go5m07REZ2HqMX7CDA38FjPRuQk+9k5e4j/Lwhg+S4cAZ2rMPMjQf4YU06AElVw9idmUutKiHMfrIbfg7daVbKk7RALgUtkJVS6gLkZcP77eFYOkQlWo9qDaHJjVCvK7iK4Iu+1ol9g76H2h3sTmyLvEInfg5z2gzlEb9s5+Up1umFEcH+3NmhDo1rRvLxL9tZuzcbgM8Gt6PrJdVsyaxURVUmBbIx5j7gQaAhEHTm+9qDrJRSlcyhbZCzH2p3OvsJe7mZMLIH5B21iuTqjcs/oxfKK3QyasEO2teLpWVildN2ivOLnHR8ZRYN4sL527WXkhgdQmz47/4vVyl1AUpaIJf4SCBjzEDgXWApEAyMBsYAR7GOn37xwqIqpZTyWVXrQ93Of3z8dGgM3P41+AXA6Kut1gxFcIAfD3WrT5s60b9rowjy9+PJqxqydGcmN76/gK6vz+FHdwuGUqp8lObMzMeAV7B2kAE+EJFBQBJwAjjs4WxKKaUqgqr1YfBPEJME4/8Euxfbncjr3d6+Nt8N7cw7A1rRoHo4j49fxbKdmXbHUqrSKE2B3AD4BXC5H4EAInIEeBl41OPplFJKVQzRdWDgZGsu8sR7rdaL4k4cgY+7wfePwdF9tkT0Ns0SoujdohYjB6ZQMyqY20cs5s//W8W+rBN2R1OqwitNgXwCcIjVtLwfa+f4pByglieDKaWUqmCCI6HfKKsA/uYea1Tc531g/bcw/R/WBIwVn8NbTWHrDLvTeo3Y8CAmPXQZN7aqxbT1+3lgzHLyi5x2x1KqQitNgbwWqO9+Pg942hjT0RjTFnge2OThbEoppSqaxLZw3RuQOhNmvgj718I398LKMXDZY/Dwcmu+8rSnoKjA7rReIzoskNdubsFbt7VkTVo2r0zZhE6hUqrslOao6Y/5bdf4GeBnYL779THgRg/mUkopVVG1uQuy98LxA9D1/+CDjlD1Euj+tHUz31UvWSfxjegOf5oE4Trq7KReTWow+LJ6jFqwg+9W7eXey5MwGG5vX5uokAC74ylVYVzwHGRjTBjQEQgFfhWRQ54MVp50zJtSStko5wAERUJA8G9rG3+ACXfDpb3h5k/sy+aFCp0uJixPY9zSPazekwVAXEQQL/dtxpWNq9ucTinv5vE5yMaYy4EVIpJzlvfCgDYi8kupk3oBLZCVUsoLzfk3zPkXtLsfrn71j0fJVVIFRS62HcihwOniqW/WsGn/MQZ1rMOzNzTRE/iU+gMlLZBL02IxG2vHeMlZ3mvkft8nDwpRSinlhbo8AXlZsOgDa/pFu/tO32Wu5AL9HTSuFQnA5GGd+fe0TXwyfwe5BU5ev6WFzemU8m2lKZDP9etoEKC31CqllPIcP3/o9S84tBVmPAO/vmsVyVunQ0w9qNYIEtpCvS52J7VdoL+DZ65vTGigH+/O2kZYkD83t0mgaXwULpcwfcN+WteOJi5Sf8FQqiTO2WJhjKnLbzfm/QwM4/fTKkKAwUALEamPD9IWC6WU8mIFx2HDZJj5AhxLh+rN4MAGECeEV4fH18ORnXB0r3XktX+g3YltU+R0cfenS5m39RAhAX6EBfkBhkM5+QT6OQj0d/Bojwbc06Uexmgbhqp8PNKDbIx5DngOKH5R8f9Fift1ETBUREZcWFx7aYGslFI+IHsvZO2C2h1h30pInQWz/glN+sL6SYBAQjuodgl0etT6byUkIuw/mseQT5cRFRKA0yV0bViNg8fyST2Yw7yth7j7srr847rG2qusKh1PFch1gLpYRfAsYCiw4YzL8oEtInLBZ2AaYxoC/yu2lAQ8C1QB7gUOutefFpEp7s/8DRiC1drxiIj85F6/Gngbqx96pIi8er6frwWyUkr5IBEYdwds/tEqjJvfCjOeA1cRGAfU6QjXv2VNyDiyE2q1gkq+ayoi/POHjYxasINWtavw6d3tdDycqlTKYopFV2D52aZYeJIxxg/YC7QH7gZyROSNM65pDHwFtMM6we9n4ORWwRbgSiANWAoMEJEzi/rTaIGslFI+SgT2rrB2i4MirNfZe2DB27B2AvgFQmRN65S+uCZQoxkkpEC7e+1ObhsR4btV+3jy69W0T4rh8Z6X0Lp2NA7dTVaVgMenWIjI3DN+wLVAYyAd+FZEckud8ux6AKkisusc/VF9gHEikg/sMMZswyqWAbaJyHZ3xnHua89ZICullPJRxkBCm9NfV6kN171p3dA36mqrOG7/AGyeAmvGwbpvrN3m4Cj7ctvIGMONreI5Uejkue/Wc/OHC6kaHki3hnGkHcklK7eQx6+8hF5NatgdVSnbnHOopDHmYWPMzDPWHMaYqcD3wGvAF8BKY4ynjjrqj7U7fNIwY8waY8woY0y0ey0e2FPsmjT32h+tK6WUqmyqNYRB31ttFle/Co+ugXtngasQfnoasnbbndBWA9rVZvkzPXm7f0s6Jlflp/X72XU4l/wiF09PXMuJAh1OpSqv801d7wtsPmNtMNALGAu0wCpo44B/XGwYY0wg0Bv42r00HEgGWmLtVL958tKzfFzOsX62n3WfMWaZMWbZwYMHz3aJUkopX1ejKaQMtnaWjYH4NtC8P6wcA++0sv5biUUEB9CnZTzvDmjFqmev4tenruD1m5tz+HgBoxbssDueUrY5X4HcCJh/xtotQCZwj4isFZHxwOvAtR7Icw3WaX0ZACKSISJOEXEBI/itjSINSCz2uQRg3znWf0dEPhaRFBFJqVbNU5vfSimlvN5NH8Fj66wb+6Y+BUfT7U7kFfwcBmMMKXVj6NWkOv/9eQvLdx2xO5ZStjhfgRyDVXQCp26g6wzMFpGCYtctwSpGL9YAirVXGGNqFnuvL7DO/Xwy0N8YE2SMqQc0cGdYCjQwxtRz70b3d1+rlFJK/aZKIvR5z2q3mHA3FObZncirvNavBbWqhDD406X8vCGDkt7Qr1RFcb4COQMo3qXfGutgkIVnXOcECrgIxphQrOkTE4stv2aMWWuMWQN0Bx4HEJH1wHism++mYc1gdopIEdZhJj8BG4Hx7muVUkqp08Umw40fwO6F8HlvyNxudyKvERUawJgh7akSGsA9ny9j2FcrtSdZVSrnm4M8AYgFrhSRImPMcOA+oImIbCp23cPAQyJyaVkHLgs65k0ppSqxdRNh8sPgLIA7J0JcYwiLtTuVVyh0uhg5bwev/bSJjkmxjLqrLcEBfnbHUuqClXTM2/l2kF8COgC7jTHrgfuBScWLY7c/AYsuKKlSSillp6Y3wbBlEJUIn/eB15Nh8cfWTOVKLsDPwYPdknnj5hb8mnpYb9xTlcY5C2QRWQVcjnUQx3asY6fvKH6NMSYeqxXjyzLKqJRSSpWtyJrQb6R1iEidy2DqX2DMTeBy2Z3MK/Rrk0DPS6vz/qxtvDp1E9m5hXZHUqpMlfgkvYpMWyyUUkqd4nLCL2/AnH/BwO8gqZvdibzCnsxc/u+bNSzekUl4kD8tEqtQJyaUdvViaJ8UQ1xEsN0RlTovjx81XZFpgayUUuo0hXnw5iWQ1B2ufgUc/hAeZ3cqr7A2LZvRv+5ga0YOOw4dJye/CIC+reJ545YW+OmR1cqLaYFcClogK6WU+p2f/g4L37Oe+4dAlyfg4EZo0Ata3GZvNi9R5HSxft9RJq3ay+gFOwn0c3BVk+q8dVtLAvzOd5uTUuWvpAWyf3mEUUoppXxOzxesiRbH9kHaMpj9krW+ZTo07g0BIfbm8wL+fg5aJFaheUIU0aGBbM44xg9r0jmSW8B7A1oTFuSPS4T3Zm3j19RDvN2/FYkxoXbHVuq8dAcZ3UFWSilVAodTYf9a+HoQ9HweOj9udyKv9PWyPfx90jqqhQeRcTSPahFBpGfnEeTvIDYskD6t4vnLVQ1xaCuGsoGnxrwppZRSCqyDRRr3gYbXwc8vwNJPIGs3zH8LTuiRzCfdkpLIyIEpHMktoF29GETg3QGt+GJIy8+AbQAAIABJREFUe6pHBTN8Tioj52/nyPGLOl9MqTKlO8joDrJSSqlSKMyD8X+CrdN/W2t4HfQfC0Z3RU8qdLp+14csIgwYsYhF2zMJ8nfwaM8GPNStPiLC9A0ZNI2PIr6Ktq6oslOuPcjGmGeBdOALEdED7ZVSSlVcAcHQ/ytY+QXkZUHeUZj/H1j/rXXoiAI46016xhg+GdSWX7Yc5LtV+3ht2mbyCl0czsln7OLdNIuPYtLQy3QShrKdR3aQjTEnJ6kfBP4jIv++6C8tR7qDrJRS6oK5nPBhZyjKg4cWgX+Q3Yl8gtMlPDF+FZNW7QOgS4OqzNt6iIhgf65pWoOm8VE0qRVJzagQaumusvKQch3zZoypA4QBnYHLReTOi/7ScqQFslJKqYuy9WcY2w/a3A03/NfuND5DRFiyI5OwIH+a1Irk84W7WL0nix/WpFPg/O0UwwHtajP4sro4RWhUI9LGxMrX6RzkUtACWSml1EWb8Rws+C/c8A60GWR3Gp+WX+Qk83gB6/ceZeH2w4xesAOXgL/DMGnoZTSNj7I7ovJRWiCXghbISimlLprLCWNvhp3zrSOqE9qBnx434Amr92SxZEcmI+ZtJ6/QyR0d6vB/Vzei0Oli+JxU2teLoX1SrN0xlQ/wSIHsvvmupERE/lmK672GFshKKaU8IjcTPu5qjX+LSoQ7voa4S+1OVWGs25vNm9M3M3vzQfq3TWRj+lFWp2UTFRLAlEe76AQMdV6eKpBdf/jm74mI+JXieq+hBbJSSimPObLLmmixaDgYBwxdBMHaEuApRU4X/T5cyJq0LBpWj+C6ZjX5cG4qiTGhfD6kHXERwXZHVF5MWyxKQQtkpZRSHrd3OYzsCYnt4cp/QmJbuxNVGPlFTgqdQniQ1cIyf+shBn+6FD+H4bGeDbi3S5Ke1KfOSk/SU0oppewU3waufR0ObYWv+lvtF8ojgvz9ThXHAJ0bVGXaY13o3KAqr0zdxFs/bwGsUXJKXQgtkJVSSqmy0vYeGDjJOop67mt2p6nQkqqF8/Gf2nBLmwTem72N+z5fRpPnpvHvaZtwaaGsSumcBbIxxmWMcZbwUVReoZVSSimfUaMZNOkLa8ZBUb7daSo0Ywwv9mlK5/pVmb4hg6Sq4Qyfk8rwual2R1M+5nzzZ14E9NcupZRS6mK0GADrJsCWn6Bxb7vTVGghgX58Mqgtq9OySKkTzSPjVvHm9M20rh1Nx2QdBadKRm/SQ2/SU0opVcacRfB2cwirCvfM0vnI5Sgnv4je780n83gBT19zKR2TY0mMCUVEGL1gJzWigvlxTTqDO9ejTZ1ou+OqMlamUyyMMeFANJApIscvIJ9X0QJZKaVUmVs3ESbcDR2HWVMtHHobUHnZdfg4Qz5bxrYDOYQF+jFiUApr07J5ZeqmU9dUiwhi6qNdqBoeZGNSVdbKZIqFMaaXMWYZkAXsBLKNMUuMMVdeWEyllFKqkmjSF1KGwML34Ken7U5TqdSJDePHRzozaehlxEeHcO9ny/j3tE1c07QGj/VswFu3teDgsXy+XpZmd1TlJUpcIBtjegE/AuHAP4GHgJeACGCKFslKKaXUORgD170J7R+ExcNh8iNQkGt3qkojyN+PlolV+GRQW/z9HNSPC+eNW1rwWM9L6NsqgRaJVZi6Lt3umMpLlLjFwhizEDgCXC8irmLrDuAHoIqIdCqTlGVMWyyUUkqVG5cTfn4efn0XLukFt3wKAXpEcnk6cCyP0ED/02Ypfzg3lVenbuK2lET6t0ukVe1oNu8/xo5Dx7m6aQ0b0ypP8ngPsjEmF7hFRH48y3vXA+NFJLTUSb2AFshKKaXK3dKR8OMTEFELLrkKrnzx9COpM3eAfxBE1rIvYyWScTSPB8csZ2tGDsfyi2iRWIWtGcfILXAy8aFOtK4djYhgjKHI6cLfT3vIfVFJC+TS3EabD0T+wXsR7veVUkopVRJt74Gql8DCD2DlWEidDfGtIbY+OAJgwdtQpTY8tNBqz1BlqnpkMBMfuoyc/CK+Wrybn9bvp0NSLOv2ZnPbRwsxxlBQ5KJaRBBZuQW0rxfLOwNaERMWaHd0VQZKs4P8LdAMuFJEdhRbrw3MANaLyE1lkrKM6Q6yUkopW22baRXER3ZA1m5rLSgK8rPhzm+gfk9781Via9OymbgyjUB/B4F+DvZk5hLo72DSqn20SqzC2Hva626yDymLFotLgAVAFLAISAdqAB2wplp0FpGtF5zYRlogK6WU8ip52eAXCO+0stosbh8P1RranUoV883yNJ74ejVVwwMZMTCFVrV1hrIv8PiYNxHZAjQH3gGCgNZAMPA20NJXi2OllFLK6wRHWTfu3TYGThyB99vBlL/Asf12J1Nu/dok8PGf2hDg5+CvE9bwwvfrycotsDuW8pAS7SAbY/yApsA+ETlY5qnKme4gK6WU8lrH9sO8/8CSj6zXKYPh2jdABA5vBWchVG+qB4/Y5PvV+3j4q5UAXNO0Bv93dSMOH8+nde1ojPaOex2Ptli4R7nlA9eJyHQP5PMqWiArpZTyejsXwIZJsORjiG8Dh7dZrRhg9SvX6wLtH4Cje6FeV4isaW/eSmTT/qPMWJ/BmzO2nFrrlBzLF0Pa4+fQItmbeHSKhYi4jDF7gLCLTqaUUkqp0qt7mfWIbWCdxNfoWmh4HSCw61dYNRY2/WBdG1ET7psLEdVtjVxZNKoRScPqEbSuE82WjGMczy/ijelb+G7VXm5qnWB3PHUBSnOT3v8B12JNsfBok40xpiHwv2JLScCzwOfu9bpYR1vfKiJHjPVvFm+78+QCd4nICvd3DQL+4f6el0Tks/P9fN1BVkop5VOcReB3xh7XtpmQsR5qtoAvb4Pk7jDgK3vyVXIul3D9u/PJK3Iy889dtdXCi5TFHOQIIBnYboyZhjXFonh1LSLyXOlinvrgZqAlnOp33gt8CzwFzBSRV40xT7lf/x9wDdDA/WgPDAfaG2NigOeAFHe25caYySJy5EJyKaWUUl7pzOIYoH4P6wFw+RMw6yXYu8KarazKlcNhuKNDbf7+7Tq2ZOTQsEaE3ZFUKZWmo/9poJb7MRj4O9ZObfGHJ/QAUkVkF9AHOLkD/Blwo/t5H+BzsSwCqhhjagK9gBkikukuimcAV3sol1JKKeUb2j8AITEw6UGYeD+sm2h3okqn56VWe8vPGzNsTqIuRIl3kEWkvG6P7Q+c/Deh6iKS7v756caYOPd6PLCn2GfS3Gt/tP47xpj7gPsAateu7bHwSimllO2CIuDWz2HszXBoK6wZB6kzIbE9+AdDUndY/SVgoNPD4Cyw5i0rj6keGUyLhCjGLtpFt4bVaFIr6vwfUl6jxAWyMSbwXL3HxpiaJ4vZC2WMCQR6A38736VnWZNzrP9+UeRj4GOwepBLEVMppZTyfvW6wMMrIDAMFvwX5r8FK8dY7508pQ9g2Sjr9L6G18ANb0NYVfsyVzDP9W7CQ2NWcOuHC/nkrrZ0SIq1O5IqodLsCn9p/qDL3N3eMNsDea4BVojIyX+PyHB/98mfccC9ngYkFvtcArDvHOtKKaVU5RMVDyFVoOfz8MQWeHw9dHrEKo77fWIdYZ29B1reDltnwIddYO9yu1NXGK1rRzN52GXUqhLCoFFLmLZuP0VOF+v2ZpNX6LQ7njqH0kyx2A98IyJDz1ivAcwB8kWkxUWFMWYc8JOIjHa/fh04XOwmvRgR+asx5jpgGNYUi/bAOyLSzn2T3nKsU/4AVgBtRCTzXD9Xp1gopZSqNEQg54A1Aq4wD3L2Q3Rd2LcK/ncnOPxh2LKz3wioLsjhnHzuGr2UtXuz8XcYilxC/bhwRt/VlsSYULvjVSoePSjE/YVtsHaJ3xSRF9xr1d1rLqCbiBy6iMChWP3DSSKS7V6LBcYDtYHdwC0ikuneyX4P6wa8XOBuEVnm/sxgrBsKAV4+WWyfixbISimlFLBpCowbAJdcbfUkF56A9vdbO83qohQUufhqyW7Ss/OoGRXMm9M3Uyc2jAkPdiTI38/ueJWGxwtk95deBXwPPAxMBOZitWl0K9YW4XO0QFZKKaWwdpfH9IP9a63WjILjkJMBvV6x2jIcAdDuXquvWV2UaevSeWDMCl7r15xb2yae/wPKI8qkQHZ/8R3AaKx+3wKs4nj/BaX0ElogK6WUUmeRdxTG3Q475/22ltAWBn0PASGnX5uxAUJjIKJG+Wb0USJCtzfmUDsmlC+GtLc7TqVR0gL5nDfpGWOSznwAC4GPgHDgXiC02HtKKaWUqiiCI+GOCdD7XXh8A9w8GtKWwrz/nH7dog9heEcYe4t1OMkJPZ/rfIwx3NC8FvO2HqLvBws4nJNvdyRVzPk68LfxB2PSsEaqzTljTZtolFJKqYokIBhaD7SeR90EW6bB/P9Ys5aDwq05y4uGW+/vXwMjukNsA2uXObKmfbl9wG1tE/ll60FW78li+JxU/nF9Y7sjKbfzFch3l0sKpZRSSvmGa9+A4wdhxjO/rVVvahXEH3W1WiwObIBPr4Vbv4AaTe3L6uUSY0KZPKwzfx6/ii8W7aJjciyNakbiZwzfrtxLg7hwejaubnfMSqnUPcgVkfYgK6WUUqXgcsGhzYCB6Dpg/MA/EApyrd7ktKVWu0VeFnQcBj1f0LFx55BxNI9bP1rIrsO5+DsMMWGBHDiWT2xYIMv+0ZM/OIZCXQCP9CArpZRSSv2OwwFxl0JcI6sg9g+01gNDwRhIbAePrISUwbDwPfj+Ecg955EElVr1yGC+ebATb93WgoY1IjiaV8gtbRI4fLyAnYdz7Y5XKZXq1zljTFdgANZc4uAz3hYR6eGpYEoppZTyYaExcP1bEFoVfnkN1n0DA8ZBcne7k3mlquFB9G2VwNVNanIkt4AThU6+Xp7GwtTD1KuqY/XKW4kLZGPM/cBw4DCwFTjzdkvd/1dKKaXU6bo/DQ2ugu8fhXF3wA3/haY3W7vQ6ndCAv0ICQxBRIiLCGLc0t2EBfkxZW06dWPDeKhbfaJCA+yOWeGV5iS9LcASYLCIFJRpqnKmPchKKaVUGTuWYR1lnbYEqjWCbk9Bk752p/JqE1ek8fdv13Gi0El0aADZJwpJrhbOkM716N4ojiU7MrmmaQ38/fSXjZIqi6OmjwO9RWTmxYbzNlogK6WUUuXAWQjrJsKCt+HAeujwEHQcClEJdifzWpnHC0jPPkFytXBW7s5i6JcryDxegJ/D4HQJD3VL5sFuyUQE665ySZRFgfwL8IWIjLjYcN5GC2SllFKqHDkLrZaLVV9aUzAeWmzNW1bn5XIJC1IP8dHc7QDM33aI8CB/Jg29jPpx4Tan835lMcXiEeAxY8zlFx5LKaWUUpWeXwDc+AH8aSIc2WntKKsScTgMXRpUY8w97Rk5KIX/3tYSP4fhbxPXUFDksjtehVGaAvl7IAGYbYw5ZozZfcZjVxllVEoppVRFlHwFNL7ROplv81TIOwo7foETWXYn8wnBAX7c2Cqe53s3ZunOI9z3xTI27T9KkVML5YtVmjFvM/njY6eVUkoppUqv18uwdTp81R+CoiA/GyJqwu3joWZzu9P5hL6tEsjJK+KlHzdy9X/nkVQ1jBf6NKFdvRiC/P3sjueT9CQ9tAdZKaWUslXGeji4CWb/C+p1hS3TrPWmN0GXJyGkir35fETG0TxmbzrAG9O3cCgnn+YJUXx5bwfCg/QUw5M8fpNeRaYFslJKKeVF9q2EiffB4W3QtB/0G2l3Ip+SfaKQaevSefrbdXRKjuWTQW0J9NdRcFDyArnUv1IYY1oADfn9SXqIyOel/T6llFJKqdPUagXDlsLc12D2yxDXGFoMgMiadifzCVEhAdzWtjYGw1+/WcOdnyzmtX7Nqasn8pVYaca8VQF+BDqcXHL/99QXiIhPNrroDrJSSinlhVxO63CRzVPALxB6/QsCwyGmHuxZAo2ug9hku1N6tW+Wp/H89+spKHLx+i0t6N2ilt2RbFUWc5A/AK4AhgDzgL5ANjAY6Aj0F5HlF5zYRlogK6WUUl6qKN+abPH9o3B0r7VmHCAuiEqEoUsgMNTejF4u42geD3+1kqU7M6kVFcKD3ZK5o31tjDHn/3AFUxYFcirwAjAWKATaniyIjTHDgTARGXjhke2jBbJSSinl5favg53zYP9ayE6D9g/AuAFw6Q1w9b8hKt7uhF7tRIGTj3/ZzoLUQyzZkUnHpFjuvbwe3RvGVapCuSwK5FzgKhGZ735+rYjMcb93JTBORGIvIrNttEBWSimlfNAvr1t9ysYBN42Axr3tTuT1RIQvl+zmzelbyDxeQIO4cK5pVpOh3ZNPjYQTEfKLXMzadIBVe7Lo3zaRpGoV45S+srhJbz9wcs7KLqy2ijnu1/VLlU4ppZRS6mJd/hdofhuMuwOm/AXq94BAvRHtXIwx3NG+DremJPLdqn2MX7aHd2Zu5ZctB3m0ZwMAXp2yic0Zx059ZtH2wzzYNZnDxwu4qXU8oYEVf2xcaXaQPwf2iMjfjTFPA88CnwFFwCBgsojcXmZJy5DuICullFI+bPdiGHUVNOkLfT8C/yC7E/mUqWvTefLr1RwvcAIQEeTP3ZfVpU3dGLJyC3h03KpT17ZMrMI9XepxXbOaPtmaURYtFslALRGZZ4wJAF4FbgNCgWnAwyJy+CIy20YLZKWUUsrHLXgHZjwD4dUhPgXSV8OgyTrlooT2Zp1g+8EcipxC/bhwEmOsGx9FhIWph4kIDmD7oRyembSOo3lF3NG+Nl0aVKNXk+o+VSjrQSGloAWyUkopVQFsnwuLhsP2OVB0AprdCv1G2J2qQnG5hCcnrGbiCmuiyKCOdXigWzLH853UjQ3F38+7DyTRArkUtEBWSimlKpgZz1q7yrePh0uusjtNhSIipGfnMXrBDkbM23FqvUmtSIZ1r8/erBPszsxlX9YJ/ByGsEB//P0Mh3MKCPR3UC0iiEKnkBAdwtDu5Xsbm8cKZGNMBNAJCADmiEiOMaYh8DzQHDgIvCMiEy86tU20QFZKKaUqmPwc+PRaOLQN7v7ROp1PedysTRls2n+MiCB//jVlEycKf+tjjo8OQQRy8osocLqoGh5EfqGTQzn5BPo7qBIayM9/7lqueT1SIBtjLgF+BuKxTs7bD9wATHW/3g4kA9FALxH5+eKjlz8tkJVSSqkK6FgGjOwJznwYuhhCou1OVKGlZ5/gwNF86sSGEhUScN7eZBEp9/7lkhbI52sU+SeQB1yFdcT0BmASsBJIFJH2QG1gLvDURSVWSimllPKkiOrQfwwcPwjfPwZH99mdqEKrGRVCi8QqVAkNLFHh6803952vQL4MeFFEZorIEuBhrN3k90QkD0BEcoF3gaZlmlQppZRSqrRqtoAuT8KGSfBeO9jyk92JlA84X4FcA0gt9vrk8zN/BUsHqnkqlFJKKaWUx1zxdxi2DGKTYMJg66hqpc7hfAWyA3AWe33y+ZmNyzoKQymllFLeq2oDuPVzEBf8+CToFC91DiUZVhdvjEkyxiQBSWeuudcTyi6iUkoppZQHRNeF7k/Dlqmw/lu70ygvVpICeQKw1f3Y5F6bVGxtK/D1xQYxxlQxxkwwxmwyxmw0xnQ0xjxvjNlrjFnlflxb7Pq/GWO2GWM2G2N6FVu/2r22zRijNw4qpZRS6jftH7RGvn3/GBzaanca5aX8z/P+3eWSwvI2ME1EbjbGBGIdYd0LeEtE3ih+oTGmMdAfaALUAn52j6QDeB+4EkgDlhpjJovIhvL6QyillFLKi/n5W60WH10OPz5hHUet1BnOWSCLyGflEcIYEwlcDtzl/rkFQME5xn/0AcaJSD6wwxizDWjnfm+biGx3f+8497VaICullFLKUqW2Ndli+t+t46mTyvewCuX9vOXA7CSsE/lGG2NWGmNGGmPC3O8NM8asMcaMMsacnPAdD+wp9vk099ofrf+OMeY+Y8wyY8yygwcPevQPo5RSSikvlzLYKpS/GgC7FtqdRnkZbymQ/YHWwHARaQUcxzp4ZDjWSX0tsUbJvem+/mxby3KO9d8vinwsIikiklKtmk6oU0oppSqVwFAYMsM6XW/GMzrVQp3GWwrkNCBNRBa7X08AWotIhog4RcQFjOC3Noo0ILHY5xOwZjP/0bpSSiml1OkiakDnxyBtqXWQiFJu57tJr1yIyH5jzB5jTEMR2Qz0ADYYY2qKSLr7sr7AOvfzycCXxpj/YN2k1wBYgrWD3MAYUw/Yi3Uj3+3l+WdRSimllA9pdSesGgsThsC6b8A4IKIW5B6GtCVw50SITbY7pSpnXlEguz0MjHVPsNiONUHjHWNMS6w2iZ3A/QAist4YMx7r5rsiYKiIOAGMMcOAnwA/YJSIrC/vP4hSSimlfERACAz6AWb9EzZMtlovts6AojwIDIfP+8AdEyCukd1JVTkyoj03pKSkyLJly+yOoZRSSilvUJQP+TlwNA3G3gLOAquIrtHU7mTqIhljlotIyvmu85YeZKWUUkop7+D//+3debyVVb3H8c8XEFCIGAQ0ERHDKUtETMxUnM0hzXkgQXG8dtObedUGu1qmNpm30kxQ0TQkR1QUCTH1miQ4K2I4goKADAZOgL/7x1oHHw8H8OA5+znD9/167dfez3rWs/fvWZvX5nfW/u31tIF2XWD9reH4sdBqbbjhUFg0O+335GKT5wTZzMzMbGU6bwzHjIL35sPwPeFXm8J9Pyo7KqtnTpDNzMzMVmW9L8PRo6BdN1j0FvzjD7DwjbKjsnrkBNnMzMxsdXrvAieMg9OfhhYt4aZjYMH01R9njZITZDMzM7NPq9NGcNi1MHca3Hl62dFYPXGCbGZmZlYbWxwAu54LL42HF+4uOxqrB06QzczMzGpruxNTbfJtp8Kcqalt4Rvw6BWwbGm5sdln1pAuFGJmZmbWOLRqDUfeCFftDtfsCz22g0Wz4M0n0hrKu5xVdoT2GXgG2czMzGxNdOwJg0dDty3gjUkpOe60MUy4EB76ddnR2WfgGWQzMzOzNdVtCxhyF7w7D157BHoPTD/eG38BtO8O2wwqO0JbA06QzczMzD6rdTrDFvunx9+6EhbPSYlyu66w6d7lxma15hILMzMzs7rUshUccT10/xLc9G145cGyI7JacoJsZmZmVtfafh4G3Qade8PIQfD2S2VHZLXgBNnMzMysPrTrAkePhBYtYPR3y47GasEJspmZmVl96dQLdj4LXnsYpj9WdjT2KTlBNjMzM6tP/QZD245w64kw4efw6v+VHZGthhNkMzMzs/rUpj0cfRN8tAz+fglcuy9c/Q2YO63syGwlvMybmZmZWX3rOQDOeBqWvAdPXA8PXAwj9oe+x8CMf8IhV0P7rmVHaZlnkM3MzMwqQYLW68D2J8OQu2GtteGhX6Vl4B4fUXZ0VuAE2czMzKzSum8J35kMZzwLvXaCySNSCYY1CE6QzczMzMrQogV03BC2PwUWvg4Tryw7IsucIJuZmZmVafP9oM9eMP58eHpU2dEYTpDNzMzMyiXBgZfDBtumpeBeGFN2RM2eE2QzMzOzsrXvCoNugS9sA7edAgteT4nyG4+XHVmz5GXezMzMzBqCtdaGQ4bD5TvAZX0hlkHL1nDYCNh837Kja1Y8g2xmZmbWUHTZBPb7NWx1MBzxZ+i+FdxyAsx+oezImhVFRNkxlK5///4xadKkssMwMzMz+6R3ZsKVOwGCI65PFxyxNSZpckT0X10/zyCbmZmZNVQd1s8XFWkLV+8D918If/sfWPhG2ZE1aa5BNjMzM2vIum4Gpz4Cd3wHHvxFanv1YRgyBlq1Lje2JsoJspmZmVlD1+ZzcMgw6LkDLPsAxp0HT98E/b5ddmRNkksszMzMzBqDlmvBgFPga9+FrpvDY8MgAt55E565GWY+lfpNHgEPXAz/nlVuvI2YZ5DNzMzMGhMJtjsBxnwfLukFS96FZR8Cgk12g5fGp35T7oSTH4QWLcuMtlFygmxmZmbW2PQfmtZNfmMytFobvnwoPHp5qk0ecBps0A9uGQoTr4Qd/qPsaBudBpMgS+oIDAO2AgI4HpgK3AT0Al4FDo+I+ZIEXAbsC7wLDImIx/PzDAZ+lJ/2ZxExooKnYWZmZlb/WrSAbQalW5VDhn38OAKeHgX3/RDarQtfObzyMTZiDakG+TLg3ojYHNgamAKcA4yPiD7A+LwN8A2gT76dBFwBIKkz8BNge+CrwE8kdarkSZiZmZmVToLDroGNdoRbT4Q/bA9P/uXj/b4Oxio1iARZUgdgZ2A4QER8GBELgAOBqhngEcBB+fGBwHWRPAp0lLQ+sDcwLiLmRcR8YBywTwVPxczMzKxhaN0OBt0CA3+QyjFuPwV+tSlc1BN+uQlMfwwWzYFZz5QdaYPTUEosegNzgGskbQ1MBk4HukfETICImCmpW+6/ATC9cPyM3Lay9hVIOok0+0zPnj3r7kzMzMzMGopWbWDg2bDTmfDE9fD6o9C2A7w4Fm44FN5fkPqd9TK061JurA1Ig5hBJiXq/YArImIbYDEfl1PURDW0xSraV2yM+FNE9I+I/l27dq1tvGZmZmaNR8tW0P84OPhK2PeXaWZ5o6/B2p3T/mdGlRtfA9NQEuQZwIyImJi3byYlzG/l0gny/exC/w0Lx/cA3lxFu5mZmZlVWbcPHPUXOPsVWL8vTLoalrxfdlQNRoNIkCNiFjBd0ma5aXfgeWA0MDi3DQbuyI9HA8cqGQAszKUYY4G9JHXKP87bK7eZmZmZWU0GngtzX4The8ADl8B7C2DZ0rKjKlVDqUEG+E/gBkmtgZeB40gJ/ChJQ4HXgcNy3zGkJd6mkZZ5Ow4gIuZJ+inwWO53QUTMq9wpmJmZmTUym+0De18ET/wZHvh5urXrBp02gq8ckS5KopqqWIEX74O5U2H7U1MZRxOh8DIf9O/fPyZNmlR2GGZmZmbleu5v/mAkAAAMqElEQVR2ePtfMOtZmPdSWuFiuxNhl7OhfVeYeg88fCm0bJ2u3vfmE+m+x1fhkKugU6+0hNziOdCu68oT65JImhwR/VfbzwmyE2QzMzOzFXz0EfztPHjkd2n7C9ukxLnDF1Ly+9HSlAD3Px7G/hCWLYEdT4cpo2H289B71zT7PPNJWLsTfOlg6LB+qafkBLkWnCCbmZmZ1SACXrofZkyCqWNSkrzn+dD285/st3AG3HM2vHAXtOkA2w6BiX9Ms8uI5YuKrdMlzTIPuTutzVxhTpBrwQmymZmZ2WcUAc/8FbptAet9Gd6dB3OmpseL3oIpd8L8V9LFSY66sZQQnSDXghNkMzMzs6bv0ybIDWKZNzMzMzOzhsIJspmZmZlZgRNkMzMzM7MCJ8hmZmZmZgVOkM3MzMzMCpwgm5mZmZkVOEE2MzMzMytwgmxmZmZmVuAE2czMzMyswAmymZmZmVmBE2QzMzMzswInyGZmZmZmBU6QzczMzMwKnCCbmZmZmRU4QTYzMzMzK1BElB1D6STNAV6r8MuuC8yt8Gs2dx7zcnjcK89jXnke83J43CuvsY/5RhHRdXWdnCCXRNKkiOhfdhzNice8HB73yvOYV57HvBwe98prLmPuEgszMzMzswInyGZmZmZmBU6Qy/OnsgNohjzm5fC4V57HvPI85uXwuFdesxhz1yCbmZmZmRV4BtnMzMzMrMAJspmZmZlZgRPkCpO0j6SpkqZJOqfseJoSSVdLmi3p2UJbZ0njJP0r33fK7ZL0v/l9eFpSv/Iib7wkbShpgqQpkp6TdHpu97jXE0ltJf1T0lN5zM/P7RtLmpjH/CZJrXN7m7w9Le/vVWb8jZ2klpKekHRX3va41yNJr0p6RtKTkiblNn++1DNJHSXdLOmF/Pm+Q3MbdyfIFSSpJfAH4BvAlsBRkrYsN6om5Vpgn2pt5wDjI6IPMD5vQ3oP+uTbScAVFYqxqVkKnBkRWwADgNPyv2mPe/35ANgtIrYG+gL7SBoAXAJcmsd8PjA09x8KzI+ILwKX5n625k4HphS2Pe71b9eI6FtYe9efL/XvMuDeiNgc2Jr0b75ZjbsT5Mr6KjAtIl6OiA+BkcCBJcfUZETEg8C8as0HAiPy4xHAQYX26yJ5FOgoaf3KRNp0RMTMiHg8P/436UN0Azzu9SaP3aK8uVa+BbAbcHNurz7mVe/FzcDuklShcJsUST2A/YBheVt43Mvgz5d6JKkDsDMwHCAiPoyIBTSzcXeCXFkbANML2zNym9Wf7hExE1IyB3TL7X4v6lj+CnkbYCIe93qVv+Z/EpgNjANeAhZExNLcpTiuy8c8718IdKlsxE3Gb4H/Bj7K213wuNe3AO6TNFnSSbnNny/1qzcwB7gmlxMNk9SOZjbuTpArq6bZA6+zVw6/F3VIUnvgFuCMiHhnVV1raPO411JELIuIvkAP0jdTW9TULd97zOuApP2B2RExudhcQ1ePe93aMSL6kb7GP03Szqvo6zGvG62AfsAVEbENsJiPyylq0iTH3QlyZc0ANixs9wDeLCmW5uKtqq968v3s3O73oo5IWouUHN8QEbfmZo97BeSvPR8g1X93lNQq7yqO6/Ixz/s/z4qlSLZ6OwLflPQqqTxuN9KMsse9HkXEm/l+NnAb6Q9Cf77UrxnAjIiYmLdvJiXMzWrcnSBX1mNAn/yr59bAkcDokmNq6kYDg/PjwcAdhfZj869vBwALq746sk8v11QOB6ZExG8Kuzzu9URSV0kd8+O1gT1Itd8TgENzt+pjXvVeHArcH75CVK1FxLkR0SMiepE+u++PiGPwuNcbSe0kfa7qMbAX8Cz+fKlXETELmC5ps9y0O/A8zWzcfSW9CpO0L2nWoSVwdURcWHJITYakvwADgXWBt4CfALcDo4CewOvAYRExLyd2vyetevEucFxETCoj7sZM0teBh4Bn+Lgu8wekOmSPez2Q9BXSD2RakiY5RkXEBZJ6k2Y2OwNPAIMi4gNJbYHrSfXh84AjI+LlcqJvGiQNBL4fEft73OtPHtvb8mYr4MaIuFBSF/z5Uq8k9SX9GLU18DJwHPnzhmYy7k6QzczMzMwKXGJhZmZmZlbgBNnMzMzMrMAJspmZmZlZgRNkMzMzM7MCJ8hmZmZmZgVOkM3MVkHSDpJGSpoh6UNJ70h6TNJPqxbNbyokDZQUhdt7kp6XdF5ec7mq3wOSHigx1JUqnMPAsmMxs8ar1eq7mJk1T5LOBH5JuhjEj0jrgbYHvgacBPQnXQK3qfku6cJG6wB7k9YU/yJwbJlBmZlVihNkM7MaSNqVlBxfFhH/VW33GEkXAYdVPrKKmBIRj+bH90vqBgyRdEZE+HLJZtbkucTCzKxmZwNz8/0KImJxRFxbbJN0vqTHJS2UNFfS/fnSq8U+VSUAB0m6UtI8SfMlXSqppaTtJD0sabGk5yTtXf21Je0iabykf+d+YyVtVa3P3pIeybEskjRV0nlrOBaP5fsv1rRTUtsc/7P5tWZJulPS5oU+2+bzPrCG46/NJSwtC20nSnpK0vt5LIdL6lztuK6SbsxlLwskXQd0XMNzNDNbzgmymVk1kloBuwDjIuLDWhy6AXApcBAwBJgNPJgvD13db4HFwBGky7SekduuA64GDiZdovhWSesWYtsPGA8sAgYBRwOfAx6StGHu0xsYDbySn/+bwG+AdrU4l6KN8/2Clexvk2P4GbAfcCrQFnhU0noAETGZlGifXDxQUkfgcGBYRCzLbRcDlwN/y7GfRbqM7T3FJBq4FdifdHnzI4ClwO/W8BzNzJZziYWZ2Yq6kBK816vvyMnzchGxtPD4hEK/lsC9wHPAUOD0ak91f0R8Lz8elxPf7wA7RcTD+TlmAk+Rks4Rue9lwN8jYvlMrKQJpProM0mJdj+gNXBqRLxT9Xqf9uSBFvk81wH2IiW8T0bEizV1joiFQPVzHwu8BRxF+qMBUtI7XNJGEfFabjs2xzosH9uLlBCfHxEXFJ7zReBh4ADgdkl7Al8HjoqIkbnbWEn3AD1qca5mZivwDLKZ2YpUY2OaDV1SvBUTZkl7SJog6W3SbOYSYFNgsxqe7p5q2y8Ai6uS40IbQNXMcB9gE+AGSa2qbsC7wD+AnXP/J/Nrj5R0aK4hro2x+fiFwF9JP1I8aFUHSDpc0kRJC0jnvpj0g8biuY8kzUKfWGg7Gbg7Imbk7T1J/zdVP8eJwDuFc9wBWAbcUi2UkZiZfUZOkM3MVjQXeB/oWUP7dvl2VXGHpH7AGFLpw1BgQO73FGk2urr51bY/pFoJQ6G8o+r4qkR3ONUSdVKpQZd83DTS6hMtgOuBWTl53WUV51x0Wo59K6B9RBxQmPFdgaQDgJuAKaSSj+3z8XMKsRMR7wPXAENz4rsTsCXwx8LTVZ3jtBrOsUPVOQLrA/MjYkm1cN76lOdoZrZSLrEwM6smIpZKehDYU1LrqkQ1l1NMApC0f7XDDiHNnB5cTNokdWLltbu19Xa+P5dUn1vd8nrpiJgATJDUBtgRuAC4W1KviJi7mtd5MSIm1SKuI4FpETGkqkHSWkDnGvpeAXwPOBD4FvAqaca6StU57sWKf0QU988EOklaq1qS3L0WcZuZ1cgJsplZzX4BjAMuAaov81aTdUhf+UdVg6TdSLPQr9RRTFNJCeWXIuLiT3NARHxAWqqtPXAH6Qd3q0uQa2sd0h8HRd8GWlbvGBEvSbqPVGfcF7ggIj4qdBkHfAT0jIhxq3jNf+TnP4RPllUcWfvwzcw+yQmymVkNImK8pHOAi/MqFNeREt22pLriI0l1tlUJ8b2kH8hdK+ma3OfHwBt1GFNIOg24Q1JrYBQp2e1OunjJ6xHxG0mnkGp1xwDTgXVJs85vAs/WVTwF9wIHSboUuAvYlnSxkZXNnF9OStaXkFbsWC4n0JcAv5e0GfB3UrnLhqT65GERMSEixkl6GLgyr/LxL9JKFp9Y7s7MbE24BtnMbCUi4hfATqSv9X9OKmu4GRhMqrntU7U0WUSMJSWFO5KSxONJKzRMq+OYxpCS33aklR/Gkma71yPNqkKqe24HXATcR1pG7hVgt4h4ry7jya4CLiQlqHeSVt04gPQjv5rcTfph4R0RMav6zoj4AelKhTuT/gi4g7Qe9XxSIlzlYNIfAReR3o9WpJVAzMw+E0XE6nuZmZnVkbxE233AHhExvux4zMyqc4JsZmYVIWkToDdpXeQPImLbkkMyM6uRSyzMzKxSfkxa//kDUvmJmVmD5BlkMzMzM7MCzyCbmZmZmRU4QTYzMzMzK3CCbGZmZmZW4ATZzMzMzKzACbKZmZmZWcH/A0EA1hkxtc85AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.subplots(figsize=(10,6))\n",
    "plt.plot(smart_bets.mean(axis=1), label='Mean Bankroll - Neural Net 2 Sizing Bets');\n",
    "plt.plot(dumb_bets.mean(axis=1), label='Mean Bankroll - Always Bet $10');\n",
    "plt.xlabel('Games Played', fontsize=16)\n",
    "plt.ylabel('Bankroll, Starts with $10,000', fontsize=16)\n",
    "plt.legend()\n",
    "plt.tight_layout()\n",
    "plt.savefig('gamble_results', bpi=150)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 288,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.subplots(figsize=(10,6))\n",
    "plt.hist(smart_bets.iloc[-1], label='Mean Bankroll - Neural Net 2 Sizing Bets', bins=20);\n",
    "plt.hist(dumb_bets.iloc[-1], label='Mean Bankroll - Always Bet $10', bins=20, alpha=0.6);\n",
    "plt.xlabel('Games Played', fontsize=16)\n",
    "plt.ylabel('Bankroll, Starts with $10,000', fontsize=16)\n",
    "plt.legend()\n",
    "plt.tight_layout()\n",
    "plt.savefig('gamble_hist', bpi=150)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 282,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.12006613622668261"
      ]
     },
     "execution_count": 282,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "smart_bets.mean(axis=1).iloc[-1]/dumb_bets.mean(axis=1).iloc[-1] - 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 223,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Y\n",
       "-1.0    0.49586\n",
       " 0.0    0.08993\n",
       " 1.0    0.41421\n",
       "Name: lose, dtype: float64"
      ]
     },
     "execution_count": 223,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model_df_smart.groupby(by='Y').count()['lose']/model_df_smart.shape[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 224,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Win or Tie Rate: 0.5041404438555813\n"
     ]
    }
   ],
   "source": [
    "print('Win or Tie Rate:', 1-model_df_smart.sum()['lose']/model_df_smart.count()['lose'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 146,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "data_smart = 1 - (model_df_smart.groupby(by='dealer_card_num').sum()['lose'] /\\\n",
    "                  model_df_smart.groupby(by='dealer_card_num').count()['lose'])\n",
    "data_random = 1 - (model_df.groupby(by='dealer_card_num').sum()['lose'] /\\\n",
    "                   model_df.groupby(by='dealer_card_num').count()['lose'])\n",
    "\n",
    "data = pd.DataFrame()\n",
    "data['smart'] = data_smart\n",
    "data['random'] = data_random\n",
    "\n",
    "fig, ax = plt.subplots(figsize=(12,6))\n",
    "ax.bar(x=data.index-0.2, height=data['smart'].values, color='blue', width=0.4, label='Smart')\n",
    "ax.bar(x=data.index+0.2, height=data['random'].values, color='red', width=0.4, label='Coin Flip')\n",
    "ax.set_xlabel(\"Dealer's Card\",fontsize=16)\n",
    "ax.set_ylabel(\"Probability of Tie or Win\",fontsize=16)\n",
    "plt.xticks(np.arange(2, 12, 1.0))\n",
    "\n",
    "plt.legend()\n",
    "plt.tight_layout()\n",
    "plt.savefig(fname='dealer_card_probs_smart', dpi=150)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 153,
   "metadata": {},
   "outputs": [],
   "source": [
    "# old = pd.read_csv('old_data_smart_dealer.csv', header=None)\n",
    "# ax = pd.concat([data_smart, old.set_index(0)], axis=1).rename({'lose': 'With Card Count Data',\n",
    "#                                                                1: 'No Card Count Data'},\n",
    "#                                                                axis=1).plot(kind='bar', figsize=(12,6));\n",
    "# ax.set_xlabel(\"Dealer's Card\",fontsize=16)\n",
    "# ax.set_ylabel(\"Probability of Tie or Win\",fontsize=16)\n",
    "# plt.tight_layout()\n",
    "# plt.savefig('card_count_comp', bpi=150)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 154,
   "metadata": {},
   "outputs": [],
   "source": [
    "# data_smart = 1 - (model_df_smart.groupby(by='player_total_initial').sum()['lose'] /\\\n",
    "#                   model_df_smart.groupby(by='player_total_initial').count()['lose'])\n",
    "# data_random = 1 - (model_df.groupby(by='player_total_initial').sum()['lose'] /\\\n",
    "#                    model_df.groupby(by='player_total_initial').count()['lose'])\n",
    "\n",
    "# data = pd.DataFrame()\n",
    "# data['smart'] = data_smart[:-1]\n",
    "# data['random'] = data_random[:-1]\n",
    "\n",
    "# fig, ax = plt.subplots(figsize=(12,6))\n",
    "# ax.bar(x=data.index-0.2, height=data['smart'].values, color='blue', width=0.4, label='Smart')\n",
    "# ax.bar(x=data.index+0.2, height=data['random'].values, color='red', width=0.4, label='Coin Flip')\n",
    "# ax.set_xlabel(\"Player's Hand Value\",fontsize=16)\n",
    "# ax.set_ylabel(\"Probability of Tie or Win\",fontsize=16)\n",
    "# plt.xticks(np.arange(4, 21, 1.0))\n",
    "\n",
    "# plt.legend()\n",
    "# plt.tight_layout()\n",
    "# plt.savefig(fname='player_hand_probs_smart', dpi=150)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 155,
   "metadata": {},
   "outputs": [],
   "source": [
    "# old = pd.read_csv('old_data_smart.csv', header=None)\n",
    "# pd.concat([data_smart, old.set_index(0)], axis=1).plot();"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 156,
   "metadata": {},
   "outputs": [],
   "source": [
    "# model_df_smart.groupby(by='player_total_initial').sum()['hit?']/model_df_smart.groupby(by='player_total_initial').count()['hit?']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 157,
   "metadata": {},
   "outputs": [],
   "source": [
    "# old_hit_prob = pd.read_csv('old_aces_only.csv', header=None)\n",
    "# old_hit_prob.set_index(0, inplace=True)\n",
    "# aces_only = model_df_smart[model_df_smart['dealer_card_num'].isin([2,3,4,5,6])]\n",
    "# hit_prob = aces_only.groupby(by='player_total_initial').sum()['hit?']/aces_only.groupby(by='player_total_initial').count()['hit?']\n",
    "# hit_prob = pd.concat([hit_prob, old_hit_prob], axis=1)\n",
    "# hit_prob.rename({'hit?': 'Hit Prob. With Card Count Data',\n",
    "#                  1: 'Hit Prob. No Card Count Data'}, axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 158,
   "metadata": {},
   "outputs": [],
   "source": [
    "# ax = hit_prob.plot(kind='bar', figsize=(12,6));\n",
    "# ax.set_xlabel(\"Player's Hand Total\",fontsize=16)\n",
    "# ax.set_ylabel(\"Probability of Hitting\",fontsize=16)\n",
    "# plt.legend(loc='lower left')\n",
    "# plt.tight_layout()\n",
    "# plt.savefig('hit_prob_comp', bpi=150)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}