Stock Price Prediction Using Time Series Models

This repository contains code for predicting stock prices using various time series models. The dataset used in this project is from NSE-TATAGLOBAL, which includes historical stock prices. The project demonstrates how to preprocess the data, check for stationarity, and apply different forecasting models including Moving Average, ARIMA, and SARIMA.

Table of Contents

• Prerequisites
• Data Import and Preprocessing
• Stationarity Check
• Model Fitting
  • Moving Average
  • ARIMA
  • SARIMA
• Usage
• License

Prerequisites

Make sure you have the following Python packages installed:

• pmdarima
• statsmodels
• prophet
• pandas-profiling
• seaborn
• numpy
• matplotlib
• sklearn

You can install these packages using pip:

pip install pmdarima statsmodels prophet pandas-profiling seaborn numpy matplotlib sklearn

Data Import and Preprocessing

1. Import Dataset

   data = pd.read_csv(r'C:\Users\admin\Downloads\NSE-TATAGLOBAL.csv')

2. Convert Date Column and Reverse Data

   data['Date'] = pd.to_datetime(data.Date, format='%Y-%m-%d')
   data = data.iloc[::-1]

3. Set Date Column as Index

   data.set_index('Date', inplace=True)

Stationarity Check

The stationarity of the time series is tested using the Augmented Dickey-Fuller (ADF) test. The series is transformed to make it stationary by applying a logarithmic transformation and differencing.

def test_stationarity(timeseries):
    # Transformation to make the data stationary
    timeseries_log = np.log(timeseries)
    timeseries_log_diff = timeseries_log.diff().dropna()
    
    # Plot rolling statistics and perform ADF test
    ...

Model Fitting

Moving Average

This section uses a moving average method for prediction. The performance is evaluated using Root Mean Squared Error (RMSE).

def moving_avg_prediction(data):
    ...

ARIMA

Auto ARIMA is used for forecasting stock prices. The model automatically selects the best parameters for ARIMA.

def arima_prediction(data):
    ...

SARIMA

The Seasonal ARIMA (SARIMA) model is used to capture seasonality in the time series. The model is fitted with parameters for both non-seasonal and seasonal components.

def sarima_prediction(data):
    ...

Usage

1. Load Data

   data = pd.read_csv('your_data.csv', parse_dates=True, index_col='Date')

2. Run Forecasting Models

   moving_avg_prediction(data)
   arima_prediction(data)
   sarima_prediction(data)

License

This project is licensed under the MIT License - see the LICENSE file for details.