Time series method for forecasting model for amount of ginger plant production

Abstract

The ginger plant is a medicinal plant with great potential to be developed as an ingredient in traditional medicine and a raw material for drinks and food. From 2015 to 2021, almost all of the crops in this rhizome group experienced an increase in harvested area, but only ginger plants experienced a decrease in production. This is due to unpredictable weather changes, which can affect crop yields. Good production results can help farmers and industry in processing ginger crops. Therefore, accurate forecasting is needed to determine the quality of decision-making. This research uses primary data from ginger crop harvest from January 2015 to December 2019. Several trials have been carried out using time series forecasting methods: Double Exponential Smoothing (DES) and Triple Exponential Smoothing (TES). The aim is to find the accuracy of several time series methods in predicting the amount of ginger production so that farmers do not fail to carry out scientific forecasting. This research shows that the best forecasting method is to use the TES forecasting model with a Mean Absolute Percentage Error (MAPE) value of 38.10% compared to DES with a MAPE value of 42.49%. This shows that the TES method is better and more capable of forecasting the amount of ginger plant production.