Forecasting of Potato Production in Bangladesh using ARIMA and Mixed Model Approach

Abstract

Listen

Translate article

A time series model is used to forecast future values by identifying patterns of historical movement of a variable. This study attempted to develop the best potato predicting model in Bangladesh using BBS provided secondary annual data on area and production of potatoes in Bangladesh from 1970–71 to 2019–20, using the most recent accessible criteria for selecting a model, such as AIC, BIC, RMSE and others. The ARIMA (0, 2, 2) model is the box-Jenkins ARIMA model with the best selection for forecasting potato output throughout Bangladesh. When considering the area of the potato, the mixed model, i.e., ARIMA (0, 2, 3), beats the univariate ARIMA (0, 2, 2) model. The mixed model's 95 percent confidence interval of the prediction value is shorter than that of the ARIMA model. As a result, the forecasting performance of the mixed model outperforms that of econometric models such as ARIMA; this might be due to the inclusion of explanatory factors such as area. The comparison of the real and predicted series demonstrates that the model used to estimate potato production in Bangladesh is statistically sound. The models forecast well at acceptable levels. As a result, depending on Bangladesh's expected potato production, these models can be used for policy objectives.