Comparison of Simple Moving Average and Least Squares Methods in Forecasting Rice Production in Padang City

Abstract

Rice production is an important agricultural indicator for supporting regional food availability and agricultural planning. Fluctuations and long-term changes in production require appropriate quantitative forecasting methods to provide information about future production levels. This study aims to compare the performance of the Simple Moving Average (SMA) and Least Squares Method (LSM) in forecasting rice production in Padang City and to determine the method that better represents the historical data. A quantitative descriptive approach was employed using annual rice production data for 2016–2025 obtained from the Kota Padang Dalam Angka publications issued by Badan Pusat Statistik. SMA generated forecasts from the average of the two preceding observations, while LSM estimated production through a linear time-trend equation. Forecasting performance was evaluated using Mean Absolute Deviation (MAD), Mean Squared Error (MSE), and Mean Absolute Percentage Error (MAPE), with all calculations performed using Microsoft Excel. The results showed that rice production generally exhibited a declining trend during the observation period. SMA produced MAD of 10.784,53 tons, MSE of 162.101.692,43, and MAPE of 17,39%, whereas LSM produced lower error measures, with MAD of 5.927,95 tons, MSE of 46.858.619,17, and MAPE of 8,14%. The 2026 forecasts generated by SMA and LSM were 46.685,00 tons and 38.451,59 tons, respectively. Based on the historical error measures, LSM provided a better representation of the declining production pattern. The findings indicate that model selection should consider the characteristics of the observed time series rather than computational simplicity alone.