prac04_regression.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Practical 4: Regression Mining\n",
    "\n",
    "### In this practical\n",
    "1. [Resuming from practical 3](#resume)\n",
    "2. [Building your first logistic regression model](#build)\n",
    "3. [Understanding your logistic regression model](#viz)\n",
    "4. [Finding optimal hyperparameters with GridSearchCV](#gridsearch)\n",
    "5. [Feature selection](#fselect)\n",
    "\n",
    "---\n",
    "\n",
    "**Written by Hendi Lie (h2.lie@qut.edu.au) and Richi Nayak (r.nayak@qut.edu.au). All rights reserved.**\n",
    "\n",
    "This practical note introduces you to regression mining in Python. Regressions offer a different approach to predictive mining as compared to decision trees. As parametric models, they assume a specific association structure between inputs and target. This relationship can generally be represented by linear mathematical formula, which is used to classify/predict the target value. In contrast, decision trees do not assume any association structure; they simply seek to isolate concentrations of cases with like-valued targets.\n",
    "\n",
    "In this practical, we will focus on logistic regression. Logistic regression is a class of regression model for classification task. It is similar in structure with linear regression, yet to make a classification prediction, transform the prediction value using logistic function. This practical will cover the key mathematical concepts of weights, logistic function and cost function of logistic regressions.\n",
    "\n",
    "You will continue to use the **Veteran** dataset that you had prepared in practical 2.\n",
    "\n",
    "## 1. Resuming from practical 3 <a name=\"resume\"></a>\n",
    "\n",
    "We will reuse code for data preprocessing from practical 2 and data partitioning from practical 3."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import classification_report, accuracy_score\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "from dm_tools import data_prep\n",
    "\n",
    "# preprocessing step\n",
    "df = data_prep()\n",
    "\n",
    "# set the random seed - consistent\n",
    "rs = 10\n",
    "\n",
    "# train test split\n",
    "y = df['TargetB']\n",
    "X = df.drop(['TargetB'], axis=1)\n",
    "X_mat = X.as_matrix()\n",
    "X_train, X_test, y_train, y_test = train_test_split(X_mat, y, test_size=0.3, stratify=y, random_state=rs)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2. Building the logistic regression model<a name=\"build\"></a>\n",
    "\n",
    "### 2.1. Standardisation\n",
    "\n",
    "Most machine learning/data mining algorithms, including regression models, are\n",
    "sensitive to input variables on different scales. Input variables on different\n",
    "scales makes comparison between data points difficult. (Recall the lecture\n",
    "example of two variables with different scaled values: student grade (in the\n",
    "range of 1-4) and student age (\\~20-65)). In addition, it also adversely affects **gradient descent** (an algorithm used for training models such as regressions, neural networks, Support Vector Machines, etc) by making weights on larger scale inputs to update much faster than smaller scale inputs, resulting in suboptimal model performance.\n",
    "\n",
    "> To read more about feature scaling, read this excellent blog post from Sebastian Raschka [link](http://sebastianraschka.com/Articles/2014_about_feature_scaling.html)\n",
    "\n",
    "To avoid this problem, it is necessary to perform standardisation on the input variables. Standardisation rescale input variables to have mean of 0 and standard deviation of 1, ensuring all of them are on the same scale. In `sklearn`, standardisation can easily be done using `StandardScaler`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Before scaling\n",
      "-------------\n",
      "Variable #0: min 0.0, max 13.0, mean 3.20 and std dev 2.13\n",
      "Variable #1: min 1.0, max 89.0, mean 10.50 and std dev 8.95\n",
      "Variable #2: min 0.0, max 9.0, mean 1.86 and std dev 1.59\n",
      "Variable #3: min 0.0, max 41.0, mean 5.58 and std dev 4.73\n",
      "Variable #4: min 0.0, max 450.0, mean 15.97 and std dev 12.07\n",
      "After scaling\n",
      "-------------\n",
      "Variable #0: min -1.505311488472091, max 4.600729037733905, mean -0.00 and std dev 1.00\n",
      "Variable #1: min -1.0623391496250907, max 8.7746107562059, mean -0.00 and std dev 1.00\n",
      "Variable #2: min -1.1659811623481273, max 4.487429288703341, mean -0.00 and std dev 1.00\n",
      "Variable #3: min -1.1795617205202689, max 7.485041166941727, mean -0.00 and std dev 1.00\n",
      "Variable #4: min -1.3225650969557856, max 35.95505089608766, mean 0.00 and std dev 1.00\n"
     ]
    }
   ],
   "source": [
    "from sklearn.preprocessing import StandardScaler\n",
    "\n",
    "# initialise a standard scaler object\n",
    "scaler = StandardScaler()\n",
    "\n",
    "# visualise min, max, mean and standard dev of data before scaling\n",
    "print(\"Before scaling\\n-------------\")\n",
    "for i in range(5):\n",
    "    col = X_train[:,i]\n",
    "    print(\"Variable #{}: min {}, max {}, mean {:.2f} and std dev {:.2f}\".\n",
    "          format(i, min(col), max(col), np.mean(col), np.std(col)))\n",
    "\n",
    "# learn the mean and std.dev of variables from training data\n",
    "# then use the learned values to transform training data\n",
    "X_train = scaler.fit_transform(X_train, y_train)\n",
    "\n",
    "print(\"After scaling\\n-------------\")\n",
    "for i in range(5):\n",
    "    col = X_train[:,i]\n",
    "    print(\"Variable #{}: min {}, max {}, mean {:.2f} and std dev {:.2f}\".\n",
    "          format(i, min(col), max(col), np.mean(col), np.std(col)))\n",
    "\n",
    "# use the statistic that you learned from training to transform test data\n",
    "# NEVER learn from test data, this is supposed to be a set of dataset\n",
    "# that the model has never seen before\n",
    "X_test = scaler.transform(X_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.2. Training logistic regression\n",
    "\n",
    "Once the input variables are scaled, we are ready to build the model. As\n",
    "discussed before, the underlying problem is classification predictive mining,\n",
    "therefore, a logistic regression model will be build for this classification\n",
    "task. In sklearn, logistic regression is implemented in\n",
    "`sklearn.linear_model.LogisticRegression`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.linear_model import LogisticRegression"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Similar to decision tree, initiate a model object (with random state to ensure consistent result) and fit it to the training dataset. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LogisticRegression(C=1.0, class_weight=None, dual=False, fit_intercept=True,\n",
       "          intercept_scaling=1, max_iter=100, multi_class='ovr', n_jobs=1,\n",
       "          penalty='l2', random_state=10, solver='liblinear', tol=0.0001,\n",
       "          verbose=0, warm_start=False)"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model = LogisticRegression(random_state=rs)\n",
    "\n",
    "# fit it to training data\n",
    "model.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "After the model is trained, evaluate the performance of the model on both training and test dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train accuracy: 0.593510324484\n",
      "Test accuracy: 0.562284927736\n",
      "             precision    recall  f1-score   support\n",
      "\n",
      "          0       0.56      0.57      0.57      1453\n",
      "          1       0.56      0.55      0.56      1453\n",
      "\n",
      "avg / total       0.56      0.56      0.56      2906\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# training and test accuracy\n",
    "print(\"Train accuracy:\", model.score(X_train, y_train))\n",
    "print(\"Test accuracy:\", model.score(X_test, y_test))\n",
    "\n",
    "# classification report on test data\n",
    "y_pred = model.predict(X_test)\n",
    "print(classification_report(y_test, y_pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The model produces an acceptable accuracy score on the test data. The training accuracy is still slightly higher than the test accuracy, which indicates slight overfitting. We will tune this logistic regression model later using GridSearchCV.\n",
    "\n",
    "## 3. Understanding your logistic regression model\n",
    "\n",
    "Take a deeper look on the model we just built. Firstly, let's go over how logistic regression models makes predictions. \n",
    "\n",
    "Recall, the task is to predict whether a person with data (input variables) $x$ will be a lapsing donor. The target is $y$, which is a binary variable where 0 = non-lapsing donor and 1 = lapsing donor. Logistic regression will predict the target variable $y$ using function $g(\\theta, x)$ as following:\n",
    "\n",
    "$$h(\\theta, x) = \\theta_{0} + \\theta_{1}x_{1} + \\theta_{2}x_{2} + \\theta_{3}x_{3} + ... + \\theta_{N}x_{N}$$\n",
    "\n",
    "$$g(\\theta, x) = \\frac{1}{1 + e^{h(\\theta, x)}}$$\n",
    "\n",
    "$$y = target$$\n",
    "$$x = input$$\n",
    "$$\\theta = weights$$\n",
    "\n",
    "$h(\\theta, x)$ is also commonly known as linear regression. The difference between logistic and linear regression is the function $g(\\theta, x)$, also called **logistic/sigmoid** function. It transforms $h(\\theta,x)$ into range of 0.0 to 1.0, which roughly translates to probability of $y = 1$ (positive, lapsing donor) given $x$. Sigmoid outputs smaller than 0.5 will be classified as 0 and the rest as 1. The output curve of $g(\\theta, x)$ looks like this:\n",
    "\n",
    "![logistic curve](http://dataminingtuts.s3.amazonaws.com/Logistic-curve.svg.png)\n",
    "\n",
    "During the training process, the model tries to learn the optimal values of $\\theta$ that results in the most accurate prediction. The algorithm used to find the optimal set of $\\theta$ is called **optimisation algorithm**. Some examples of optimisation algorithm including gradient descent (most common and simple), LBFGS, adam, sag and newton-cg. These algorithms aim to minimise the **cost function**, which is the measure of how good predictions made by a model to the supplied $y_{train}$. Given weights of $\\theta$, training set $X$ and $y$ with $m$ data points, the cost function $J(\\theta)$ of a logistic regression is as follows:\n",
    "\n",
    "$$J(\\theta) = -\\frac{1}{m}[\\sum_{i=1}^{m} y^{(i)} \\log h_{\\theta}(x^{(i)}) + (1-y^{(i)}) \\log (1-h_{\\theta}(x^{(i)}))]$$\n",
    "\n",
    "Once the model is trained, all of its weights are stored in `.coef_` array of the model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[  8.55294819e-02   5.72905765e-02   9.60262361e-02  -7.68268037e-02\n",
      "   -5.03317638e-02  -3.99429982e-02   7.37564612e-02  -5.93071375e-02\n",
      "   -1.68332723e-01   2.80447649e-01  -1.39864836e-01   5.06139983e-02\n",
      "    1.97820127e-01   4.77841914e-02   9.49498242e-02  -4.38512559e-01\n",
      "    1.10047871e-01   5.51011404e-02   2.70004445e-02   1.25856209e-01\n",
      "    4.53598078e-02  -2.50445045e-02  -2.30582374e-02   7.76478322e-02\n",
      "   -2.66611375e-02  -7.17073881e-03   2.82414971e-02   5.45915187e-05\n",
      "    4.95292495e-02   2.50351269e-02  -5.18782525e-02   2.46976545e-02\n",
      "    2.03661586e-03   4.34903010e-03  -1.69243519e-04  -1.34790895e-03\n",
      "   -2.41957340e-02   1.54527250e-02   2.44627356e-02  -5.96265462e-04\n",
      "    9.41790072e-03   1.79343011e-02  -3.78651082e-02  -1.01814670e-02\n",
      "   -1.30084774e-02   1.83655032e-02  -7.87823668e-03   8.94038392e-04\n",
      "   -3.89726384e-04   1.77516004e-02   1.19268189e-02   2.45853350e-02\n",
      "   -5.11888957e-02   1.11627368e-02  -4.77942174e-02  -1.00081318e-02\n",
      "   -3.03296908e-03   3.13872418e-02  -1.14767292e-03  -1.78770954e-02\n",
      "    3.76352505e-02   1.77136706e-02   2.15441246e-02   4.20433112e-02\n",
      "   -3.26689324e-02   4.23613050e-02  -1.14826888e-02  -2.38149054e-02\n",
      "   -4.50399335e-02   4.15010922e-02  -5.28358478e-02  -8.82769886e-03\n",
      "   -2.79017387e-02  -1.95714531e-02   7.06573697e-03  -1.79368199e-02\n",
      "   -3.52741210e-02   2.68294262e-02  -3.67792906e-02   8.25559246e-03\n",
      "   -2.19038475e-02  -2.22317776e-02   5.20675074e-03   1.87282270e-03\n",
      "   -1.53757147e-02]]\n"
     ]
    }
   ],
   "source": [
    "print(model.coef_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Each of these values corresponds to a variable/feature in our dataset. We could print out feature name associated with each coefficient with this code below."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "GiftCnt36 : 0.0855294818905\n",
      "GiftCntAll : 0.0572905765277\n",
      "GiftCntCard36 : 0.0960262360759\n",
      "GiftCntCardAll : -0.076826803721\n",
      "GiftAvgLast : -0.0503317638337\n",
      "GiftAvg36 : -0.0399429982013\n",
      "GiftAvgAll : 0.0737564612393\n",
      "GiftAvgCard36 : -0.0593071374998\n",
      "GiftTimeLast : -0.168332722532\n",
      "GiftTimeFirst : 0.280447648513\n",
      "PromCnt12 : -0.139864836229\n",
      "PromCnt36 : 0.0506139982924\n",
      "PromCntAll : 0.197820127191\n",
      "PromCntCard12 : 0.0477841914496\n",
      "PromCntCard36 : 0.0949498242078\n",
      "PromCntCardAll : -0.438512558857\n",
      "StatusCatStarAll : 0.110047871434\n",
      "DemAge : 0.0551011404084\n",
      "DemHomeOwner : 0.0270004444639\n",
      "DemMedHomeValue : 0.125856209357\n"
     ]
    }
   ],
   "source": [
    "feature_names = X.columns\n",
    "coef = model.coef_[0]\n",
    "\n",
    "# limit to 20 features, you can comment the following line to print out everything\n",
    "coef = coef[:20]\n",
    "\n",
    "for i in range(len(coef)):\n",
    "    print(feature_names[i], ':', coef[i])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Recall again that a linear/logistic regression model assumes linear correlation between input features and prediction values. This relationship is reflected in the weight values.\n",
    "* Positive coefficient means positive change in the input feature has positive correlation to the prediction value. **DemAge** has positive coefficients, which means older donors (larger age value) are predicted to be more likely to donate back.\n",
    "\n",
    "* Negative coefficient does the reverse. For example, **GiftTimeLast** has negative coefficient, thus if a donor has not donated for a while, the model predicts it is less likely for them to be a lapsing donor.\n",
    "\n",
    "In addition to understanding positive/negative correlations between features and prediction values, it is important to learn how much a feature impact the overall prediction value (i.e. feature importance). One way to answer this question is by looking at the absolute value of coefficients. Changes in an important variable (either positive or negative) should correlate to a larger impact to prediction value, thus the coefficient assigned to this variable will have a large absolute value. \n",
    "\n",
    "This method only applies when the features are standard scaled. Without scaling, each feature can be on different range of values, making larger ranged feature more important by default. For example, in a dataset with **age** and **income**, a **+10** increase in age is significant, but it is meaningless in income. In this scenario, without scaling the coefficients do not help in finding important variables.\n",
    "\n",
    "The following code will sort the coefficients by their largest absolute value and print the corresponding feature names."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "PromCntCardAll : -0.438512558857\n",
      "GiftTimeFirst : 0.280447648513\n",
      "PromCntAll : 0.197820127191\n",
      "GiftTimeLast : -0.168332722532\n",
      "PromCnt12 : -0.139864836229\n",
      "DemMedHomeValue : 0.125856209357\n",
      "StatusCatStarAll : 0.110047871434\n",
      "GiftCntCard36 : 0.0960262360759\n",
      "PromCntCard36 : 0.0949498242078\n",
      "GiftCnt36 : 0.0855294818905\n",
      "StatusCat96NK_E : 0.0776478322061\n",
      "GiftCntCardAll : -0.076826803721\n",
      "GiftAvgAll : 0.0737564612393\n",
      "GiftAvgCard36 : -0.0593071374998\n",
      "GiftCntAll : 0.0572905765277\n",
      "DemAge : 0.0551011404084\n",
      "DemCluster_47 : -0.0528358478151\n",
      "DemCluster_10 : -0.0518782524943\n",
      "DemCluster_30 : -0.0511888956977\n",
      "PromCnt36 : 0.0506139982924\n"
     ]
    }
   ],
   "source": [
    "# grab feature importances from the model and feature name from the original X\n",
    "coef = model.coef_[0]\n",
    "feature_names = X.columns\n",
    "\n",
    "# sort them out in descending order\n",
    "indices = np.argsort(np.absolute(coef))\n",
    "indices = np.flip(indices, axis=0)\n",
    "\n",
    "# limit to 20 features, you can leave this out to print out everything\n",
    "indices = indices[:20]\n",
    "\n",
    "for i in indices:\n",
    "    print(feature_names[i], ':', coef[i])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Based on this output, the most important variables in this model is PromCntCardAll, GiftTimeFirst, and PromCntAll. These important variables will be different from one model to another."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4. Finding optimal hyperparameters with GridSearchCV\n",
    "\n",
    "Similar to the previous practical, this section will guide you to tune the default model to perform better. In this example, we will only tune one hyperparameter from our logistic regression model, the regularisation strength used in the model. Regularisation is a technique used to prevent overfitting in regression models. We will not go in depth to explain how regularisation works, you can find more information here:\n",
    "* [Basic of regularisation](https://www.quora.com/What-is-regularization-in-machine-learning)\n",
    "* [L1 vs L2 regularisation](http://www.chioka.in/differences-between-l1-and-l2-as-loss-function-and-regularization/)\n",
    "\n",
    "In `sklearn` logistic regression, regularisation is implemented in the hyperparameter **C**, which denotes the inverse of regularisation strength. Smaller C means stronger regularisation. Typical values for C range from $10^{-6}$ to $10^{4}$, increasing in logarithmic order, which are the values we will use in this example.\n",
    "\n",
    "**Tips:** `GridSearchCV` can be slow for searching over a large set of possible values. To speed up the searching process, `GridSearchCV` has parallel running capability, where you can specify how many parallel processes running in the same time with `n_jobs` (-1 means GridSearchCV will use as many cores as possible)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train accuracy: 0.595132743363\n",
      "Test accuracy: 0.565381968341\n",
      "             precision    recall  f1-score   support\n",
      "\n",
      "          0       0.56      0.59      0.57      1453\n",
      "          1       0.57      0.55      0.56      1453\n",
      "\n",
      "avg / total       0.57      0.57      0.57      2906\n",
      "\n",
      "{'C': 0.001}\n"
     ]
    }
   ],
   "source": [
    "# grid search CV\n",
    "params = {'C': [pow(10, x) for x in range(-6, 4)]}\n",
    "\n",
    "# use all cores to tune logistic regression with C parameter\n",
    "cv = GridSearchCV(param_grid=params, estimator=LogisticRegression(random_state=rs), cv=10, n_jobs=-1)\n",
    "cv.fit(X_train, y_train)\n",
    "\n",
    "# test the best model\n",
    "print(\"Train accuracy:\", cv.score(X_train, y_train))\n",
    "print(\"Test accuracy:\", cv.score(X_test, y_test))\n",
    "\n",
    "y_pred = cv.predict(X_test)\n",
    "print(classification_report(y_test, y_pred))\n",
    "\n",
    "# print parameters of the best model\n",
    "print(cv.best_params_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "GridSearchCV returns `C` = 0.001 as the optimal parameter. The model has a slight (almost negligble) improvement on the test data. In the next section, we will explore model improvement techniques beyond GridSearch."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "## 5. Input transfomation and dimensionality reduction\n",
    "\n",
    "Other than fine tuning models by performing grid search over their hyperparameters, there are many methods to improve prediction quality from a model. One way is to improve the input features itself, such as with input transformation. In this technique, methods that changes input distribution such as log transformation or exponential function are applied.\n",
    "\n",
    "Another method to improve prediction quality from a model is to perform dimensionality reduction or feature selection on the input set. The aim of dimensionality reduction is to reduce the size of input set hoping that it will improve model performance, reduce overfitting and enhance visualiation. Dimensionality reduction is divided into two processes:\n",
    "* Feature selection: Process of selecting a subset of relevant features/variables to be used in constructing models.\n",
    "* Feature extraction: Process of transforming high-dimensional feature space into lower dimension. Typically performed by finding principle components of the feature space.\n",
    "\n",
    "In the following sections, we will explore some input transformation and dimensionality reduction techniques.\n",
    "\n",
    "### 5.1. Logarithmic transformation\n",
    "\n",
    "Regression models are sensitive to extreme or outlying values in the input space. Inputs with highly skewed or kurtotic distributions are often selected over inputs with better overall predictions. To avoid this problem, analysts often regularise the input distributions using a simple transformation. This approach could improves model performance, at cost of increased difficulty in interpretation.\n",
    "\n",
    "Examine the variables with **GiftAvg** and **GiftCnt** in their names. Plot their distributions using `sns.distplot`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAl8AAAJNCAYAAAAPoRCNAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nSUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0xaca3208>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "def plot_skewed_columns(df):\n",
    "    # setting up subplots for easier visualisation\n",
    "    f, axes = plt.subplots(2,4, figsize=(10,10), sharex=False)\n",
    "\n",
    "    # gift avg plots\n",
    "    sns.distplot(df['GiftAvg36'].dropna(), hist=False, ax=axes[0,0])\n",
    "    sns.distplot(df['GiftAvgAll'].dropna(), hist=False, ax=axes[0,1])\n",
    "    sns.distplot(df['GiftAvgCard36'].dropna(), hist=False, ax=axes[1,0])\n",
    "    sns.distplot(df['GiftAvgLast'].dropna(), hist=False, ax=axes[1,1])\n",
    "\n",
    "    # gift cnt plots\n",
    "    sns.distplot(df['GiftCnt36'].dropna(), hist=False, ax=axes[0,2])\n",
    "    sns.distplot(df['GiftCntAll'].dropna(), hist=False, ax=axes[0,3])\n",
    "    sns.distplot(df['GiftCntCard36'].dropna(), hist=False, ax=axes[1,2])\n",
    "    sns.distplot(df['GiftCntCardAll'].dropna(), hist=False, ax=axes[1,3])\n",
    "\n",
    "    plt.show()\n",
    "    \n",
    "plot_skewed_columns(df)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The **GiftAvg** and **GiftCnt** columns show some degree of skew in their distribution. To normalise this skewed distributions, use the log transformation. Log transformation measure the order of magnitude of the underlying measure instead of the measure itself.\n",
    "\n",
    "Use `pandas.apply()` to apply `np.log()` transformation over these columns."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAlEAAAJPCAYAAABcoIE1AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0xc755be0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import numpy as np\n",
    "\n",
    "# list columns to be transformed\n",
    "columns_to_transform = ['GiftAvg36', 'GiftAvgAll', 'GiftAvgCard36', 'GiftAvgLast',\n",
    "                        'GiftCnt36', 'GiftCntAll', 'GiftCntCard36', 'GiftCntCardAll']\n",
    "\n",
    "# copy the dataframe\n",
    "df_log = df.copy()\n",
    "\n",
    "# transform the columns with np.log\n",
    "for col in columns_to_transform:\n",
    "    df_log[col] = df_log[col].apply(lambda x: x+1)\n",
    "    df_log[col] = df_log[col].apply(np.log)\n",
    "\n",
    "# plot them again to show the distribution\n",
    "plot_skewed_columns(df_log)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "After transformation, the columns are much more normally distributed. The next step is to test whether this transformation benefits performance of the model.\n",
    "\n",
    "Start by rescaling and resampling the training and test data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": true,
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# create X, y and train test data partitions\n",
    "y_log = df_log['TargetB']\n",
    "X_log = df_log.drop(['TargetB'], axis=1)\n",
    "X_mat_log = X_log.as_matrix()\n",
    "X_train_log, X_test_log, y_train_log, y_test_log = train_test_split(X_mat_log, y_log, test_size=0.3, stratify=y_log, \n",
    "                                                                    random_state=rs)\n",
    "\n",
    "# standardise them again\n",
    "scaler_log = StandardScaler()\n",
    "X_train_log = scaler_log.fit_transform(X_train_log, y_train_log)\n",
    "X_test_log = scaler_log.transform(X_test_log)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "After the training and test data is ready, run a grid search to find the optimal **C** value again."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train accuracy: 0.59395280236\n",
      "Test accuracy: 0.566758430833\n",
      "             precision    recall  f1-score   support\n",
      "\n",
      "          0       0.56      0.58      0.57      1453\n",
      "          1       0.57      0.55      0.56      1453\n",
      "\n",
      "avg / total       0.57      0.57      0.57      2906\n",
      "\n",
      "{'C': 0.001}\n"
     ]
    }
   ],
   "source": [
    "# grid search CV\n",
    "params = {'C': [pow(10, x) for x in range(-6, 4)]}\n",
    "\n",
    "cv = GridSearchCV(param_grid=params, estimator=LogisticRegression(random_state=rs), cv=10, n_jobs=-1)\n",
    "cv.fit(X_train_log, y_train_log)\n",
    "\n",
    "# test the best model\n",
    "print(\"Train accuracy:\", cv.score(X_train_log, y_train_log))\n",
    "print(\"Test accuracy:\", cv.score(X_test_log, y_test_log))\n",
    "\n",
    "y_pred = cv.predict(X_test_log)\n",
    "print(classification_report(y_test_log, y_pred))\n",
    "\n",
    "# print parameters of the best model\n",
    "print(cv.best_params_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Similar to the previous model with optimal **C** value (0.001), log transformation only managed to increased the model performance slightly on the test performance."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.2. Recursive feature elimination\n",
    "\n",
    "The next method is a dimensionality reduction technique called recursive feature elimination (RFE). RFE works by first training the model on all features. Each feature then will be assigned an weight, and features with smallest weight (less important) is eliminated, making a smaller feature set. This process is repeated a number of times until reaching the optimal performance.\n",
    "\n",
    "In this practical, we will use RFE with cross validation. Cross validation allows RFE to generalise better over the training dataset. RFE with CV is implemented in sklearn under `sklearn.feature_selection.RFECV`. Initiate the RFE with a logistic regression estimator and 10-fold validation and fit it to start eliminating features."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Original feature set 85\n",
      "Number of features after elimination 40\n"
     ]
    }
   ],
   "source": [
    "from sklearn.feature_selection import RFECV\n",
    "\n",
    "rfe = RFECV(estimator = LogisticRegression(random_state=rs), cv=10)\n",
    "rfe.fit(X_train, y_train) # run the RFECV\n",
    "\n",
    "# comparing how many variables before and after\n",
    "print(\"Original feature set\", X_train.shape[1])\n",
    "print(\"Number of features after elimination\", rfe.n_features_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The RFE+CV process output 40 features as the optimal number of features. This means that according to the model, there are only 40 variables in the input set that are important to the analysis. We can now `.transform()` the original input set, taking only the important features."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "X_train_sel = rfe.transform(X_train)\n",
    "X_test_sel = rfe.transform(X_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Run another `GridSearchCV` and test if the new input set improves the model performance."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train accuracy: 0.593510324484\n",
      "Test accuracy: 0.56882312457\n",
      "             precision    recall  f1-score   support\n",
      "\n",
      "          0       0.57      0.58      0.57      1453\n",
      "          1       0.57      0.56      0.56      1453\n",
      "\n",
      "avg / total       0.57      0.57      0.57      2906\n",
      "\n",
      "{'C': 100}\n"
     ]
    }
   ],
   "source": [
    "# grid search CV\n",
    "params = {'C': [pow(10, x) for x in range(-6, 4)]}\n",
    "\n",
    "cv = GridSearchCV(param_grid=params, estimator=LogisticRegression(random_state=rs), cv=10, n_jobs=-1)\n",
    "cv.fit(X_train_sel, y_train)\n",
    "\n",
    "# test the best model\n",
    "print(\"Train accuracy:\", cv.score(X_train_sel, y_train))\n",
    "print(\"Test accuracy:\", cv.score(X_test_sel, y_test))\n",
    "\n",
    "y_pred = cv.predict(X_test_sel)\n",
    "print(classification_report(y_test, y_pred))\n",
    "\n",
    "# print parameters of the best model\n",
    "print(cv.best_params_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The RFE also managed to only slightly improve the accuracy of our model on test data. More importantly, with much smaller feature set, the training, testing and prediction process is speed up significantly.\n",
    "\n",
    "RFE could also be combined with other dimensionality reduction and transformation techniques. For example, we could combine it with log transformation by running RFE on the log transformed input set. Use the following code and see the combined approach improves the model performance."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Original feature set 85\n",
      "Number of features after elimination 34\n",
      "Train accuracy: 0.592330383481\n",
      "Test accuracy: 0.570543702684\n",
      "             precision    recall  f1-score   support\n",
      "\n",
      "          0       0.57      0.58      0.57      1453\n",
      "          1       0.57      0.57      0.57      1453\n",
      "\n",
      "avg / total       0.57      0.57      0.57      2906\n",
      "\n",
      "{'C': 0.01}\n"
     ]
    }
   ],
   "source": [
    "# running RFE + log transformation\n",
    "rfe = RFECV(estimator = LogisticRegression(random_state=rs), cv=10)\n",
    "rfe.fit(X_train_log, y_train_log) # run the RFECV on log transformed dataset\n",
    "\n",
    "# comparing how many variables before and after\n",
    "print(\"Original feature set\", X_train_log.shape[1])\n",
    "print(\"Number of features after elimination\", rfe.n_features_)\n",
    "\n",
    "# select features from log transformed dataset\n",
    "X_train_sel_log = rfe.transform(X_train_log)\n",
    "X_test_sel_log = rfe.transform(X_test_log)\n",
    "\n",
    "# init grid search CV on transformed dataset\n",
    "params = {'C': [pow(10, x) for x in range(-6, 4)]}\n",
    "cv = GridSearchCV(param_grid=params, estimator=LogisticRegression(random_state=rs), cv=10, n_jobs=-1)\n",
    "cv.fit(X_train_sel_log, y_train_log)\n",
    "\n",
    "# test the best model\n",
    "print(\"Train accuracy:\", cv.score(X_train_sel_log, y_train_log))\n",
    "print(\"Test accuracy:\", cv.score(X_test_sel_log, y_test_log))\n",
    "\n",
    "y_pred_log = cv.predict(X_test_sel_log)\n",
    "print(classification_report(y_test_log, y_pred_log))\n",
    "\n",
    "# print parameters of the best model\n",
    "print(cv.best_params_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "RFE also managed to improve performance of the model on the log transformed dataset."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.3. Feature selection using another model\n",
    "\n",
    "The last method we will explore in this practical is \"select from model\". In this technique, we utilise machine learning models with ability to find feature importance and select the feature set using the computed importance from the trained model. Typically, decision tree or support vector machine models are used in this method. Recall from the previous practical that decision trees can compute feature importance internally through the tree structure. Therefore, we will use decision tree as \"selecting model\" here.\n",
    "\n",
    "We will utilise the log transformed dataset for this purpose. Initialise a GridSearchCV with DecisionTreeClassifier and fit it to the training data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'criterion': 'gini', 'max_depth': 5, 'min_samples_leaf': 50}\n"
     ]
    }
   ],
   "source": [
    "from sklearn.tree import DecisionTreeClassifier\n",
    "\n",
    "# similar parameters with the last practical\n",
    "params = {'criterion': ['gini', 'entropy'],\n",
    "          'max_depth': range(2, 7),\n",
    "          'min_samples_leaf': range(20, 60, 10)}\n",
    "\n",
    "cv = GridSearchCV(param_grid=params, estimator=DecisionTreeClassifier(random_state=rs), cv=10)\n",
    "cv.fit(X_train_log, y_train_log)\n",
    "\n",
    "print(cv.best_params_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Once the decision tree is trained, analyse the feature importance from this trained model using `analyse_feature_importance()` function we wrote in the previous practical."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "GiftCnt36 : 0.323720289918\n",
      "DemMedHomeValue : 0.168703994888\n",
      "GiftAvgLast : 0.137150279435\n",
      "GiftTimeLast : 0.0721689578918\n",
      "StatusCatStarAll : 0.0462480759406\n",
      "GiftCntAll : 0.0452643106608\n",
      "GiftCntCardAll : 0.0444400763448\n",
      "PromCntCardAll : 0.0383528300839\n",
      "DemPctVeterans : 0.0296061018056\n",
      "PromCnt36 : 0.0181470953664\n",
      "GiftAvgAll : 0.0173729693106\n",
      "StatusCat96NK_A : 0.0168629570627\n",
      "GiftTimeFirst : 0.0141290969041\n",
      "GiftAvgCard36 : 0.0132096933543\n",
      "PromCntCard12 : 0.00872994822258\n",
      "PromCnt12 : 0.00589332280972\n",
      "DemCluster_13 : 0.0\n",
      "DemCluster_10 : 0.0\n",
      "DemCluster_11 : 0.0\n",
      "DemCluster_12 : 0.0\n"
     ]
    }
   ],
   "source": [
    "from dm_tools import analyse_feature_importance\n",
    "\n",
    "# analyse feature importance from the tuned decision tree against log transformed X\n",
    "analyse_feature_importance(cv.best_estimator_, X_log.columns)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The output shows there are only 16 features with importance value more than 0. This means according to the tuned decision tree, there are only 16 important features in our dataset. We could now use this decision tree to perform feature selection.\n",
    "\n",
    "Import `SelectFromModel` module from `sklearn.feature_selection`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(6780, 14)\n"
     ]
    }
   ],
   "source": [
    "from sklearn.feature_selection import SelectFromModel\n",
    "\n",
    "# use the trained best decision tree from GridSearchCV to select features\n",
    "# supply the prefit=True parameter to stop SelectFromModel to re-train the model\n",
    "selectmodel = SelectFromModel(cv.best_estimator_, prefit=True)\n",
    "X_train_sel_model = selectmodel.transform(X_train_log)\n",
    "X_test_sel_model = selectmodel.transform(X_test_log)\n",
    "\n",
    "print(X_train_sel_model.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The shape of X_train shows there are only 14 feature left, exactly what the decision tree suggests. Next, train and tune another logistic regression model from this new data set and see if it improves the performance."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train accuracy: 0.581710914454\n",
      "Test accuracy: 0.570543702684\n",
      "             precision    recall  f1-score   support\n",
      "\n",
      "          0       0.57      0.58      0.58      1453\n",
      "          1       0.57      0.56      0.57      1453\n",
      "\n",
      "avg / total       0.57      0.57      0.57      2906\n",
      "\n",
      "{'C': 0.01}\n"
     ]
    }
   ],
   "source": [
    "params = {'C': [pow(10, x) for x in range(-6, 4)]}\n",
    "\n",
    "cv = GridSearchCV(param_grid=params, estimator=LogisticRegression(random_state=rs), cv=10, n_jobs=-1)\n",
    "cv.fit(X_train_sel_model, y_train_log)\n",
    "\n",
    "print(\"Train accuracy:\", cv.score(X_train_sel_model, y_train_log))\n",
    "print(\"Test accuracy:\", cv.score(X_test_sel_model, y_test_log))\n",
    "\n",
    "# test the best model\n",
    "y_pred = cv.predict(X_test_sel_model)\n",
    "print(classification_report(y_test_log, y_pred))\n",
    "\n",
    "# print parameters of the best model\n",
    "print(cv.best_params_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The test accuracy result shows improvement over the original feature set as well. In addition, the feature set now only has 14 features compared to the original 85, translating to huge computation power saved. This demonstrates the effectiveness of dimensionality reduction.\n",
    "\n",
    "## End notes\n",
    "\n",
    "This lecture note taught us how to build, tune and explore the structure of\n",
    "logistic regression models. We also explored transformation and dimensionality\n",
    "reduction techniques to overcome skewed distribution and reduce the size of the feature set and improve performance of our logistic regression model."
   ]
  }
 ],
 "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": "3.6.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}