Time Series Forecasting for Potato Price Prediction in Bangladeshi Market: A Comparative Study

Potatoes play a crucial role in Bangladesh's economy, offering economic benefits, versatile uses, and an alternative to rice—the daily consumed product across the country. However, the Bangladeshi market faces challenges such as seasonal price fluctuations, complex demand-supply dynamics, and unscrupulous practices by intermediaries, which adversely affect potato prices and harm both farmers and consumers. Developing an accurate forecasting model for seasonal potato prices could empower policymakers to implement interventions that safeguard the interests of farmers and consumers, thereby promoting sustainable potato production. To address this issue, four forecasting models, namely Seasonal Naïve, Damped Holt-Winter’s, ETS, and Seasonal Auto-regressive Integrated Moving Average (SARIMA) models were developed in this study, utilizing average monthly potato price data from 2012 to 2021. Leveraging average monthly potato price data, the model fitting residuals were analyzed, and error metrics were evaluated. The comparative analysis showed that except for the Seasonal Naïve model, all the models exhibit a satisfactory fit to the data and produced reasonably precise forecasts. Among the models considered, the Damped Holt-Winter's model showed the most accurate forecasts with the least mean absolute scaled error (MASE) of 0.117, followed by ETS and SARIMA. The outcomes of the forecasting models offer valuable insights for policymakers to stabilize potato prices, support farmers, and protect consumers' interests in the Bangladeshi market, fostering sustainable growth in the potato sector.