A comparative analysis of the accuracy of forecasting methods in predicting strategic food production in East Java

Abstract

Food security is one of the main pillars of sustainable national development, especially in East Java, a region that contributes significantly to the national rice production. However, data from 2016 to 2024 show a downward trend in rice production. This contrasts with the relatively stable consumption demand and poses a risk to future food stability. This study aims to predict future food needs and determine the most accurate forecasting method by comparing the naive method, moving average, single exponential smoothing (SES), and double exponential smoothing (DES) methods. The research data includes annual rice production and consumption volumes in East Java over a nine-year period. We evaluated the forecasting accuracy using the mean absolute deviation (MAD), mean squared error (MSE), and mean absolute percentage error (MAPE). The results of the analysis show that the double exponential smoothing method (with α = 0.9 and β = 0.1) provides the best performance, with the lowest error rate (MAPE) of 1.020%. This value is much more accurate than those of the naive method (6.397%), moving average method (6.359%), and single exponential smoothing method (6.530%), which are less responsive to downward trends in the data. Therefore, the DES method is recommended as the most appropriate forecasting model to assist the government of East Java with strategic planning and food security policies.