Modelling of Jowar production in Karnataka state - A hybrid approach

Abstract:

The current study used both linear and nonlinear time series techniques to forecast Jowar production from 1962-1963 to 2024-2025. The Autoregressive Integrated Moving Average (ARIMA) model was first used, and the best fit was chosen using diagnostic measures. Advanced machine learning techniques like Time Delay Neural Network (TDNN), Nonlinear Support Vector Regression (NLSVR), and their hybrid combinations of ARIMA-TDNN and ARIMA-NLSVR were used to look for any nonlinear patterns in the data. The adoption of hybrid models was validated by the BDS test on ARIMA residuals, which verified the existence of nonlinearity. RMSE, MAE and MAPE were used to assess the model performance for the nonlinear and hybrid techniques. In terms of forecast accuracy, the ARIMA (3,1,1)-TDNN (4-7-1) hybrid model outperformed the other models. Forecasts for up to 2030-2031 were also generated using the model, which predicted that Karnataka will produce 648.72 thousand tons of Jowar in 2030-31.