Trend model for forecasting sustainable potato production in Bangladesh

Abstract

Bangladesh, predominantly an agricultural nation, deeply used its culture with farming practices. Among its staple crops, potatoes hold significant importance, following rice and wheat. Every year there is a notable increase in potato cultivation and consumption, reflecting the vegetable’s growing popularity. The increasing popularity of potatoes presents a significant opportunity to enhance Bangladesh’s socioeconomic development with minimal investment required. This research aims to assess various trend models to understand how effectively they capture changes in potato production over time. Secondary data from FAOSTAT spanning from 1971 to 2021, the coefficient of determination (R²), and adjusted R² were employed to evaluate the performance of seven trend models. Notably, the quadratic, cubic, and S-curve models yielded the highest R² and adjusted R² values, indicating their superior performance in tracking potato production trends. Furthermore, to determine the accuracy of fitted models, we utilized metrics like Mean Absolute Percentage Error (MAPE) and Mean Absolute Deviation (MAD). The quadratic model, with a MAPE of 20.93% and MAD value of 514185.63, emerges as the most suitable option after rigorous statistical diagnostics. The ARIMA (0,2,2) model exhibits the most optimal correlation for forecasting potato production within the framework of Bangladesh. Through a comparative analysis between the projected dataset and the original data, it is evident that the fitted model ARIMA (0,2,2) outperforms statistically others. This implies that the model possesses the capacity to precisely forecast potato production in Bangladesh for the upcoming decade. These findings will provide critical insights for stakeholders and policymakers in developing sustainable potato production strategies.

Ann. Bangladesh Agric. 28(1): 111-125