Evaluation of statistical models for forecasting of area, production and productivity of total Pulses in Chhattisgarh

Abstract

The study has been carried out to forecast time series data from 2001 to 2023 of total Pulses production in Chhattisgarh along with the evaluation of Statistical Models for forecasting of area, production and productivity of total Pulses in Chhattisgarh. To explain the fluctuation in area, production, and productivity of total Pulses in Chhattisgarh. An effort has been made to fit linear, polynomial, ARIMA models and in order to develop a methodology that can accurately explain the variation in area, production and productivity. In addition to the model selection for future forecasts, the significant higher coefficient of determination (R2) and the lowest MAPE and RMSE values were used as model selection criteria. In this study production & productivity increased from 2001 to 2023, while the area of total pulses exhibited a fluctuating constant and dropping tendency during the predicted period which is up to five years. The prediction performance or better accuracy fitted models results in terms of MAE, RMSE, and R2 for ARIMA. It is found that ARIMA(1,2,1) models for Area, ARIMA(0,0,0) for production and ARIMA(0,1,1) for productivity , where Area with lowest MAE (53.03), RMSE (65.26) value and greater R2 (0.60) and Production with lowest MAE (64.580), RMSE (101.47) value and greater R2 (0.80) and Productivity with lowest MAE (73.14), RMSE (78.98) value and greater R2 (0.87) has outperformed than other model respectively. The future projection of the total Pulses crop showed a declining tendency for area and production whereas a minute inclining trend in productivity.