app.py

from flask import Flask, render_template, request
import pandas as pd
import numpy as np
import tensorflow as tf
import yfinance as yf
import pickle

app = Flask(__name__)

# Define tickers and load models
tickers = ['AMZN', 'TSLA', 'AAPL']
models = {}
for ticker in tickers:
 model = tf.keras.models.load_model(f'model_{ticker}.h5')
 models[ticker] = model

with open('/code/min_max.pickle', 'rb') as handle:
 min_max_scaling = pickle.load(handle)

# Function to prepare the data for model input
def prepare_data(data):
 # Assuming the data is a 1D array of closing prices
 # Reshape the data to have the shape (batch_size, timesteps, features)
 data = np.array(data)
 data = data.reshape(1, data.shape[0], 1)
 return data

# Function to get the last 60 days data for a ticker
def get_last_60_days_data(ticker):
 # Define the end date as yesterday
 end_date = pd.Timestamp.today() - pd.Timedelta(days=1)
 # Define the start date as 120 days before the end date
 start_date = end_date - pd.Timedelta(days=120)

 # Fetch the stock data using yfinance
 stock_data = yf.download(ticker, start=start_date, end=end_date, progress=False)

 # Ensure we have enough data (at least 60 days)
 if len(stock_data) < 60:
 return None

 # Extract the last 60 days 'Close' prices from the stock data
 last_60_days_data = stock_data['Close'].tolist()[-60:]

 last_60_days_data = (last_60_days_data - min_max_scaling[ticker][0])/(min_max_scaling[ticker][1] - min_max_scaling[ticker][0])

 return last_60_days_data.tolist()

# Function to predict the next day closing value using the model
def predict_next_day(ticker, data):
 model = models[ticker]
 data = prepare_data(data)
 prediction = model.predict(data)
 return prediction[0]

def scale_back_data(data,ticker):
 data = np.array(data)
 data = data * (min_max_scaling[ticker][1] - min_max_scaling[ticker][0]) + min_max_scaling[ticker][0]
 return data.tolist()

# @app.route('/')
# def hello_world():
# return "Hello World"

# Flask route to handle the main page
@app.route('/', methods=['GET', 'POST'])
def index():
 if request.method == 'POST':
 selected_ticker = request.form['ticker']
 last_60_days_data = get_last_60_days_data(selected_ticker)
 last_60_days_data_original = get_last_60_days_data(selected_ticker)
 predictions = predict_next_day(selected_ticker, last_60_days_data)
 # for _ in range(10):
 # next_day_prediction = predict_next_day(selected_ticker, last_60_days_data)
 # predictions.append(next_day_prediction)
 # last_60_days_data.append(next_day_prediction)
 # last_60_days_data.pop(0)
 predictions = scale_back_data(predictions,selected_ticker)
 last_60_days_data_original = scale_back_data(last_60_days_data_original,selected_ticker)
 return render_template('index.html', tickers=tickers, selected_ticker=selected_ticker, predictions=predictions, last_60_days_data=last_60_days_data_original)
 else:
 return render_template('index.html', tickers=tickers, selected_ticker=None, predictions=None, last_60_days_data=None)

if __name__ == '__main__':
 app.run(host='0.0.0.0',port = 7860)