Application of Univariate Time Series Models for Forecasting Area, Production, and Productivity of Aman Rice in Jalpaiguri, West Bengal, India

Abstract

Agricultural production is reliant on modern technology and historical information to enhance the present outcomes and ensure future sustainability. In this study, the area, production, and productivity of Aman rice in Jalpaiguri district using data from 1977-2022 is modelled by two popular time series modelling techniques i.e., the Autoregressive Integrated Moving Average (ARIMA) method and the Generalized Autoregressive Conditional Heteroscedasticity (GARCH) method. A comparison of the models based on the lowest values of Root Mean Square Error (RMSE), Mean Absolute Percentage Error (MAPE), and Mean Absolute Error (MAE) revealed that ARIMA models performed better in the training period on all three series. However, in the test period, the GARCH models on area and production performed better while the ARIMA model performed better on productivity. The best-fitted models selected were AR (1) - GARCH (1,1) on Aman rice area, AR (1) - GARCH (1,1) on Aman rice production and ARIMA (0,1,1) on Aman rice productivity. Using the chosen models, forecasts are produced for the subsequent ten years.