Forecasting the Financial Pulse: A Comparative Study of Classical, Probabilistic, and Ensemble Time Series Models
This study investigates the performance of three distinct time series forecasting models—ARIMA, Prophet, and Random Forest—in predicting stock prices using historical data from the State Bank of India (SBIN.NS). The data was preprocessed, tested for stationarity using the Augmented Dickey-Fuller (ADF) test, and visualized to understand underlying patterns. Results indicate varying degrees of suitability for each model depending on the nature of the data and the forecasting horizon. While ARIMA and Prophet are well-suited for capturing temporal dependencies and seasonality, Random Forest offers flexibility in handling non-linear relationships and exogenous variables.
Keywords: Time Series Forecasting, ARIMA, Prophet, Random Forest, Stock Price Prediction, Machine Learning, Financial Analysis
