Navigating the Interest Rate Forecasting Landscape: Unveiling ARIMA’s Power and Pitfalls in Comparison to Advanced Machine Learning Models

This research provides an in-depth assessment of the accuracy and efficiency of the regression and neural network models in predicting the interest rate of the Reserve Bank of India and Federal Reserve of America. It evaluates the models based on metrics such as Mean Absolute Error, Mean Squared Error, and R2 score. The results shows that the neural network model outperforming the regression models in predicting one of the banks' interest rates, while the other model fell short of expectations. The study also identified issues with the ARIMA model, which was found to be susceptible to outliers and necessitated certain parameters. Consequently, methods for improvement are proposed to address these limitations, though the accuracy of predictions cannot be assured. This research offers valuable insights for policy makers, investors, and finance analysts to gain insight into the performance of various models in predicting interest rates.