การพยากรณ์ราคารายเดือนของพืชสวนด้วยวิธีการพยากรณ์แบบเฉพาะและวิธีลำดับชั้นแบบบนลงล่าง

Abstract

Management in agricultural supply chains, the price factor is an important consideration since it influences the control of production costs. Being able to accurately predict the price in advance will result in more effective planning. The monthly forecasts of three types of horticultural crops, rubber, oil palm, and large dried coconut, were studied using an individual forecast and top-down hierarchical forecasting, and the forecasts were compared using the mean absolute error percentage and tracking signal. By dividing the data into 2 sets that are the training dataset which is monthly price data from 1999 to 2020 and the testing dataset, which is monthly price data in 2021, compared to actual price. The results showed that the rubber price was forecasted using a Damped Trend Non-Seasonal, oil palm, and large copra prices were based on top-down hierarchical forecasting based on the proper weighting. The mean absolute percentage error of the training data ranged from 6.25% to 7.03%. The comparison of the forecasted price and the real price in 2021, the mean absolute percentage error ranged from 8.91% to 9.55%. After analyzing the tracking signal values, the monthly prices were within the defined range [-6, +6]. Forecast Forecasted results are highly accurate forecasts. This work could be used as a planning tool by farmers and industrial enterprises.