Participatory Price Forecasting of Green Brinjal in Phetchabun Province

Abstract

The objective of this research was to compare and select the best forecasting time series method for forecasting the price of green brinjal in Phetchabun province in participation with farmers. The five comparison forecasting methods include: (1) Naive method (2) Moving average method (3) Simple exponential smoothing method (4) Double exponential smoothing method and (5) Winter’s exponential smoothing method. A forecast model using the selling price data of green brinjal from Talaadthai since 1 January 2021 to 30 March 2025 amount 1,550 data, then compared with the forecast values obtained from the model and the actual values. This research appropriate forecasting methods from the lowest were considered of mean percentage of absolute error (MAPE), mean absolute deviation (MAD) and mean square error (MSE).

 The results showed that (1) Winter’s exponential smoothing method (Alpha = 1.0, Gamma = 0.2, Delta = 0.2 to 1.0)  suitable for forecasting the lowest selling price of green brinjal, which gives MAPE = 0.0167 MAD = 0.0025 and MSE = 0.0019 (2) Winter’s exponential smoothing method (Alpha = 0.8, Gamma = 0.4, Delta = 0.2 to 1.0) suitable for forecasting the highest selling price of green brinjal, which gives MAPE = 0.0133 MAD = 0.0023 and MSE = 0.0021 and (3) Most farmers chose to use the Naïve method rather than the method that provides the highest accuracy, namely the Holt–Winters method, because the Naïve method is easy to understand and helps with immediate decision-making. This finding indicates that, although the Holt–Winters method provides better statistical accuracy, real world acceptance depends on translating forecasts into formats that are simple and practically usable.