Stock-Price-Prediction

This program predicts stock prices using the ARIMA,SARIMA,PROPHET,LSTM,CHRONOS,LINEAR REGRESSION models. It also detects and removes outliers from the stock price data to improve the accuracy of predictions.

#Goal This program shows that LSTM model and Chronos Model fits best for stock price prediction as time series dataset.

To implement stock price prediction using multiple models such as LSTM, ARIMA, Prophet, SARIMA, Linear Regression, Ridge Regression, and a Deterministic Model, you need to approach each method with its specific configuration and implementation. Below is an overview of how you can combine these models and their implementation steps. Steps to Implement

1. Load Stock Price Data: Use a single dataset for training all models.
2. Preprocessing: Handle missing values, feature scaling, and train-test split.
3. Modeling: o Implement ARIMA and SARIMA for time series modeling. o Use LSTM for deep learning-based time series forecasting. o Use Linear Regression and Ridge Regression for standard regression techniques. o Implement Prophet for time-series forecasting using trend decomposition. o Implement Deterministic models using moving averages or other simple statistical techniques.

import pandas as pd import numpy as np import matplotlib.pyplot as plt from sklearn.metrics import mean_squared_error, mean_absolute_error from sklearn.linear_model import LinearRegression, Ridge from statsmodels.tsa.arima.model import ARIMA from statsmodels.tsa.statespace.sarimax import SARIMAX from fbprophet import Prophet import tensorflow as tf from tensorflow.keras.models import Sequential from tensorflow.keras.layers import Dense, LSTM

Fit ARIMA model

arima_model = ARIMA(train_data, order=(5,1,0)).fit() arima_pred = arima_model.forecast(steps=len(test_data))

Evaluate ARIMA

mse_arima = mean_squared_error(test_data, arima_pred)

Fit SARIMA model

sarima_model = SARIMAX(train_data, order=(1,1,1), seasonal_order=(1,1,1,12)).fit() sarima_pred = sarima_model.forecast(steps=len(test_data))

Evaluate SARIMA

mse_sarima = mean_squared_error(test_data, sarima_pred)

Prepare data for Prophet

prophet_data = close_prices.reset_index().rename(columns={'Date': 'ds', 'Close': 'y'}) prophet_model = Prophet() prophet_model.fit(prophet_data[:train_size])

Predict

future = prophet_model.make_future_dataframe(periods=len(test_data)) prophet_pred = prophet_model.predict(future)

Evaluate Prophet

mse_prophet = mean_squared_error(test_data['Close'], prophet_pred['yhat'][-len(test_data):])

Prepare data for LSTM (reshaping for 3D tensor input)

train_scaled = train_data.values.reshape((train_data.shape[0], 1, 1))

Build LSTM model

lstm_model = Sequential() lstm_model.add(LSTM(50, input_shape=(train_scaled.shape[1], train_scaled.shape[2]))) lstm_model.add(Dense(1)) lstm_model.compile(optimizer='adam', loss='mse')

Train LSTM

lstm_model.fit(train_scaled, train_data.values, epochs=100, batch_size=32, verbose=1)

Predict using LSTM

lstm_pred = lstm_model.predict(test_data.values.reshape((test_data.shape[0], 1, 1)))

Evaluate LSTM

mse_lstm = mean_squared_error(test_data, lstm_pred)

Prepare data for Linear Regression

X_train = np.arange(len(train_data)).reshape(-1, 1) y_train = train_data.values

X_test = np.arange(len(train_data), len(train_data) + len(test_data)).reshape(-1, 1)

Train Linear Regression model

lr_model = LinearRegression() lr_model.fit(X_train, y_train)

Predict using Linear Regression

lr_pred = lr_model.predict(X_test)

Evaluate Linear Regression

mse_lr = mean_squared_error(test_data, lr_pred)

Train Ridge Regression model

ridge_model = Ridge(alpha=1.0) ridge_model.fit(X_train, y_train)

Predict using Ridge Regression

ridge_pred = ridge_model.predict(X_test)

Evaluate Ridge Regression

mse_ridge = mean_squared_error(test_data, ridge_pred)

Simple moving average model

window = 5 deterministic_pred = test_data.rolling(window=window).mean().dropna()

Evaluate Deterministic Model

mse_deterministic = mean_squared_error(test_data[window-1:], deterministic_pred)