2024, Vol. 9, Special Issue 1

The Present study has been conducted to compare the forecasting of Autoregressive Integrated Moving Average (ARIMA) model and Artificial Neural Network (ANN) model for the finger millet production in Odisha. The Production data of finger millet from 1971-72 to 2020-21 have been collected from Department of Agriculture and Farmers empowerment, Govt of Odisha. The models were fitted using the 90% of dataset and the other 10% dataset have been used for cross validation. Different models have been identified and based on the lowest value of Root Mean Square Error and Mean Absolute Percentage Error; the most efficient forecasting model have been selected. The study finds that the Neural Network Autoregressive (NNAR) model is the best fitted model due to the lowest value of Root Mean Square Error (RMSE) and mean absolute percentage Error (MAPE). The best fitted model NNAR (5, 4) is used to forecast the finger millet production for the upcoming years in Odisha. From this study, it is found that the production of finger millet will follow both increasing and decreasing trends in future years. The Production expected to decrease in 2021-22 then it will increase in 2022-23 again it follows decreasing trend in 2023-24 and increasing trend in 2024-25. it will expect to decrease in 2025-26.