long_term_performance_plotter.py

# %%
import pandas as pd
import pickle
from matplotlib import pyplot
import torch
from progress.bar import Bar

# Load the gpu (in my case it actually runs slower so i turned it off)
if torch.cuda.is_available():
    device = torch.device('cuda:0')
else:
    device = torch.device('cpu')
torch.no_grad()

    
# Load the csv data as a pandas dataframe
eurusd_stock_df = pd.read_csv('datasets/EURUSDX.csv')
test_data = eurusd_stock_df[['Open','Close','High','Low']][-1000:]

norm_test_data = ((test_data-test_data.min())/(test_data.max()-test_data.min()))

norm_test_data = torch.tensor(norm_test_data.values)
norm_test_data = norm_test_data.to(device)
norm_test_data = norm_test_data.float()

# Utility function
def unnormalize(x, field = 'Open'):
    return (x*(test_data[field].max()-test_data[field].min())) + test_data[field].min()

f = open('nets/long_term_learn/net9.obj', 'rb')
net = pickle.load(f)
net.to(device)


batch_size = 20
look_ahead_size = 5
real_open_prices = []
predicted_prices = []

net.eval()
epochs = len(norm_test_data)//look_ahead_size - 2*(batch_size-look_ahead_size)
b = Bar('Generating...',max=epochs)
for epoch in range(epochs):
    test_index = 1 + epoch*look_ahead_size
    for i in range(batch_size-look_ahead_size):
        inpt = norm_test_data[test_index+i]
        output = net(inpt)

    inpt = output[0]

    for batch_num in range(look_ahead_size):
        output = net(inpt)
        inpt = output[0]
        real_open_prices.append(unnormalize(float(norm_test_data[test_index+batch_size-look_ahead_size+batch_num][0])))
        predicted_prices.append(unnormalize(float(output[0][0])))

    net.clean()
    b.next()


pyplot.plot(real_open_prices, label='Real Prices')
pyplot.plot(predicted_prices, label= 'Predicted Prices')
pyplot.legend()
pyplot.xlabel('Days')
pyplot.ylabel('EUR / USD')
pyplot.title('SMLSTM 20 Long term (5 days) Predicitons')
pyplot.grid(True)
pyplot.show()