LSTM model building

LSTM.py

import numpy as np
import pandas as pd
from sklearn import preprocessing
from sklearn.model_selection import train_test_split

# Reading and converting Date
data = pd.read_csv('AAPL.csv')
data['Date'] = pd.to_datetime(data['Date'], infer_datetime_format=True)
data.info()

# feature selection from the dataset is 'Adj Close' because it is the closing price after adjustments for all applicable splits and dividend dist
y = data['Adj Close']

# normalize the data before model fitting, it will boost the performance (in Neural Networks). + transform 
from sklearn.preprocessing import MinMaxScaler
scaler=MinMaxScaler(feature_range=(0,1))  # scale of the output and input in the range 0–1 to match the scale of the layer of LSTM
y=scaler.fit_transform(np.array(y).reshape(-1,1)) # reshape:convert the univariate 1D array into 2D 

##splitting dataset into train and test split
training_size=int(len(y)*0.65)
test_size=len(y)-training_size
train_data,test_data=y[0:training_size,:],y[training_size:len(y),:1]

# check size
training_size,test_size
# Output:  (1937, 1043) 

# building the input variable
def create_dataset(dataset, time_step=1):
	dataX, dataY = [], []
	for i in range(len(dataset)-time_step-1):  # Time Series setps (0-99,100-200,,,,) any steps
		a = dataset[i:(i+time_step), 0]   
		dataX.append(a)
		dataY.append(dataset[i + time_step, 0])
	return numpy.array(dataX), numpy.array(dataY)

#  Initialize steps(this example 100 steps). It means that the model makes predictions based on the last 100 data 
time_step = 100
X_train, y_train = create_dataset(train_data, time_step) # apply create_dataset()
X_test, ytest = create_dataset(test_data, time_step)

print(X_train.shape), print(y_train.shape)
#Output: (1836, 100) #The 100 is time_step
#Output: (1836,)

# reshape train & input-output pairs
X_train =X_train.reshape(X_train.shape[0],X_train.shape[1] , 1)
X_test = X_test.reshape(X_test.shape[0],X_test.shape[1] , 1)