Forecasting Bulk Cooking Oil Prices for 20 District/Cities in North Sumatra Using Autoregressive Integrated Moving Average (ARIMA) Method

Abstract

Abstract— Frying is a popular method of food preparation in Indonesian society, whether it's for home-cooked meals or snacks bought outside. In Indonesia, there is bulk cooking oil available without a brand, which is cheaper. As a result, the usage of bulk cooking oil is higher compared to branded cooking oil. According to data published by BPS (Central Statistics Agency) in 2021, the per capita consumption of cooking oil in Indonesia reached 0.393L/Capita/week. The consumption of cooking oil in North Sumatra is also relatively high, but the poverty rate in North Sumatra is still high, with a figure of 1,268.19 thousand people.
 ARIMA method has advantages where the forecasting follows the data pattern and is flexible with relatively high accuracy. It is a quick and simple process (Hutasuhut, 2014). In ARIMA, there are two ways to determine the model, which are using the Auto Arima function and analyzing the ACF & PACF plots. The data used is the bulk cooking oil price data from January 2021 to January 2023 in Excel format, obtained from the Main Commodity Price System of North Sumatra.
 Based on the conducted research, the accuracy of the models generated by the Auto Arima function and the ACF & PACF plots are relatively similar. However, in some cases, the models generated by the Auto Arima function are not significant, although they have higher accuracy than the models generated from the ACF & PACF plots. The highest accuracy level is in Nias Utara district with an accuracy of 99.67% using the ARIMA(3,1,5) model, while the lowest accuracy level is in Nias Barat district with an accuracy of 75.26% using the ARIMA(2,0,0) model.
 Index Terms— ARIMA, Bulk Cooking Oil, Auto Arima, ACF&PACF Plot.