Model Analysis of Gated Recurrent Unit for Multivariate Rice Price Forecasting

Abstract

Food security, especially in the agricultural sector in the form of food price stability of rice as a national food ingredient is a strategic issue for Indonesia. Rice price forecasting is needed to mitigate rising rice food prices. Rice price fluctuations can be caused by internal factors such as bad weather or external factors such as the low selling price of rice, resulting in losses for farmers. This study aims to carry out multivariate rice price forecasting in DKI Jakarta by involving rice prices, weather, economic, and health factors using the Gated Recurrent Unit (GRU) algorithm where the accuracy test is based on the MAPE value between forecasting results and actual data. As a result of the GRU algorithm for multivariate rice price forecasting, the MAPE for training and testing is 0.964% and 2.628%, indicating that all models in the measurement category are very well represented.