predictFrontend.py

#import packages
import pandas as pd
import numpy as np
import sys
import json
import argparse
import datetime as dt
import requests
from datetime import timedelta
#importing required libraries
from sklearn.preprocessing import MinMaxScaler
from keras.models import Sequential, model_from_json, load_model
from keras.layers import Dense, Dropout, LSTM 

#for normalizing data
from sklearn.preprocessing import MinMaxScaler
scaler = MinMaxScaler(feature_range=(0, 1))

import os
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'

#lines = sys.stdin.readlines()
#ticker = json.loads(lines[0])

# ====================== Loading Data from Alpha Vantage ==================================
lines = sys.stdin.readlines()
ticker = json.loads(lines[0])

api_key = "LBG3YII29JOX2SSU"

#ticker = "AAPL"

# JSON file with all the stock market data for AAL from the last 20 years
url_string = "https://www.alphavantage.co/query?function=TIME_SERIES_DAILY&symbol=%s&outputsize=full&apikey=%s"%(ticker,api_key)

# Save data to this file
file_to_save = 'stock_market_data-%s.csv'%ticker

# If you haven't already saved data,
# Go ahead and grab the data from the url
# And store date, low, high, volume, close, open values to a Pandas DataFrame


response = requests.get(url_string)
data = json.loads(response.text)
# extract stock market data
data = data['Time Series (Daily)']
df = pd.DataFrame(columns=['Date','Low','High','Close','Open'])
for k,v in data.items():
    date = dt.datetime.strptime(k, '%Y-%m-%d')
    data_row = [date.date(),float(v['3. low']),float(v['2. high']),
                float(v['4. close']),float(v['1. open'])]
    df.loc[-1,:] = data_row
    df.index = df.index + 1
   
	

# If the data is already there, just load it from the CSV


#print the head
df.head()

#setting index as date
df['Date'] = pd.to_datetime(df.Date,format='%Y-%m-%d')
df.index = df['Date']

#plot
#plt.figure(figsize=(16,8))
#plt.plot(df['Close'], label='Close Price history')

#
#LSTM Model Start
#
#creating dataframe
data = df.sort_index(ascending=True, axis=0)
new_data = pd.DataFrame(index=range(0,len(df)),columns=['Date', 'Close'])
for i in range(0,len(data)):
	new_data['Date'][i] = data['Date'][i]
	new_data['Close'][i] = data['Close'][i]

#setting index
new_data.index = new_data.Date
new_data.drop('Date', axis=1, inplace=True)

#creating train and test sets
dataset = new_data.values

train = dataset[len(dataset)-3000:,:]
valid = dataset[987:,:]

#converting dataset into x_train and y_train
scaler = MinMaxScaler(feature_range=(0, 1))
scaled_data = scaler.fit_transform(dataset)

x_train, y_train = [], []
for i in range(60,len(train)):
	x_train.append(scaled_data[i-60:i,0])
	y_train.append(scaled_data[i,0])
x_train, y_train = np.array(x_train), np.array(y_train)

x_train = np.reshape(x_train, (x_train.shape[0],x_train.shape[1],1))

file_name = 'model_%s.h5'%ticker

if(os.path.isfile(file_name) is False):
	# create and fit the LSTM network
	print(json.dumps("Error: Model has not been trained yet."))

else:
	train = dataset[:,:]
	model = load_model(file_name)
	#predicting test here
	"""inputs2 = new_data[len(train)-60:len(train)].values

	for i in range(0,60):
		inputsNorm2 = inputs2.reshape(-1,1)
		inputsNorm2 = scaler.transform(inputs2)

		X_test2 = []
		for i in range(0,inputsNorm2.shape[0]):
			X_test2.append(inputsNorm2[i,0])
		X_test2 = np.array(X_test2)
		X_test2 = np.reshape(X_test2, (1,60,1))

		closing_price = model.predict(X_test2)
		closing_price = scaler.inverse_transform(closing_price)
		#add new data to the input set and remove the beginning data pt in the input set
		nextDataToAdd = closing_price[0]
		dataset = np.vstack([dataset, nextDataToAdd])
		inputs2 = dataset[len(dataset)-60:]
		print(closing_price)"""

	inputs2 = new_data[len(train)-60:len(train)].values
	trainInputs2 = new_data[len(train)-61:len(train)].values

	predictVals = 20

	for i in range(0,predictVals):
		inputsNorm2 = inputs2.reshape(-1,1)
		inputsNorm2 = scaler.transform(inputs2)

		X_test2 = []
		#use 60 to predict 1
		X_test2.append(inputsNorm2[:, 0])
		X_test2 = np.array(X_test2)
		X_test2 = np.reshape(X_test2, (X_test2.shape[0],X_test2.shape[1],1))

		closing_price = model.predict(X_test2)
		closing_price = scaler.inverse_transform(closing_price)
		#add new data to the input set and remove the beginning data pt in the input set
		nextDataToAdd = closing_price[0]
		inputs2 = inputs2[1:]
		inputs2 = np.vstack([inputs2, nextDataToAdd])
		trainInputs2 = trainInputs2[1:]
		trainInputs2 = np.vstack([trainInputs2, nextDataToAdd])

	new_data2 = pd.DataFrame(index=range(0,predictVals+1),columns=['Date', 'Close'])
	j = len(train)-1
	new_data2['Date'][0] = data['Date'][j]
	new_data2['Close'][0] = data['Close'][j]
	for i in range(1,predictVals+1):
		new_data2['Date'][i] = data['Date'][j] + timedelta(days=i+1)
		new_data2['Close'][i] = inputs2[i+39,0]
		

	#setting index
	new_data2.index = new_data2.Date
	new_data2.drop('Date', axis=1, inplace=True)

	new_data2['Date'] = new_data2.index.strftime('%Y-%m-%d')
	new_data2['Close'] = new_data2['Close'].apply(lambda x: round(x, 2))
	json_valid = new_data2.to_json(orient='records')
	print(json_valid)