Application of Machine Learning Techniques Models for Forecasting of Redgram Prices of Andhra Pradesh, India

Abstract

Recent advancements in Machine Learning (ML) had proven highly effective in modeling time series data, consistently outperforming traditional time series models in forecasting accuracy according to empirical studies. However, the application of ML techniques in forecasting agricultural commodity prices in India was remains scarce, despite their demonstrated success in other domains. The present study endeavours to investigate the efficiency of various machine learning (ML) algorithms, including Artificial Neural Network (ANN), Support Vector Regression (SVR) and Random Forest (RF) models, alongside traditional linear time series models such as SARIMA and GARCH models in forecasting of the monthly price series of redgram in Andhra Pradesh, India. The findings of this study indicated that the Random Forest (RF) model exhibited superior performance compared to other machine learning techniques and univariate time series models in forecasting redgram monthly prices in Andhra Pradesh. However, the forecasting accuracies of alternative techniques, including Support Vector Regression (SVR), Artificial Neural Network (ANN), GARCH, and SARIMA models, fell short of expectations. In this research, the superiority of various models was substantiated through accuracy metrics such as Mean Squared Error (MSE) and Root Mean Squared Error (RMSE). Additionally, the Diebold-Mariano test is conducted to assess significant differences in predictive accuracy among the models. The DM test also concluded that the RF model outperformed than the other models under consideration.