Some Advanced Time Series Models for Forecasting Groundnut Production in India
K Sri Yuva Surya Prakash
*
S.V. Agricultural College, Tirupati, Acharya N.G. Ranga Agricultural University, Andhra Pradesh, India.
B Ramana Murthy
Department of Statistics and Computer Applications, S.V. Agricultural College, Tirupati, Acharya N.G. Ranga Agricultural University, Andhra Pradesh, India.
Shaik Nafeez Umar
Department of Statistics and Computer Applications, S.V. Agricultural College, Tirupati, Acharya N.G. Ranga Agricultural University, Andhra Pradesh, India.
K. N. Ravi Kumar
Department of Agricultural Economics, S.V. Agricultural College, Tirupati, Acharya N.G. Ranga Agricultural University, Andhra Pradesh, India.
G. Mohan Naidu
Department of Statistics and Computer Applications, S.V. Agricultural College, Tirupati, Acharya N.G. Ranga Agricultural University, Andhra Pradesh, India.
*Author to whom correspondence should be addressed.
Abstract
Groundnut (Arachis hypogaea L.) is a major oilseed crop in India, characterized by significant production variability due to climatic and other external factors, making accurate forecasting essential for effective agricultural planning and policy decisions. This study analyses annual groundnut production data in India from 1950-51 to 2024-25 using advanced time series models and machine learning models like ARIMA, Artificial Neural Networks (ANN), Wavelet-ARIMA and Wavelet-ANN. Model performance was evaluated using MAE, MSE, and RMSE, along with diagnostic tests such as the Box-Pierce and Diebold-Mariano tests. The results reveal that the ARIMA (0,1,2) model provides superior out- in of-sample forecasting accuracy compared to ANN Wavelet-ARIMA and Wavelet-ANN. Forecasts indicate a gradual increase in groundnut production from 2025-26 to 2027-28, with increasing uncertainty over time. Overall, the findings highlight the robustness of ARIMA models in agricultural forecasting and provide valuable insights for policymakers and stakeholders informed decision-making.
Keywords: ARIMA, wavelet, forecasting, groundnut production, time series models