Forecasting the Automobile and Parts Product Export Values using Time Series Analysis

Abstract

This research aims to study the suitable time series analysis to forecast the automobile and parts product export values over the next 12 months. The time series data source gathers from the Government Open Data of Thailand official website during January 2013 to December 2021, 108 months in total. The experiment starts with creation, comparison, selection, verification, and forecasting. Time series analysis has considered five methods: Trend Analysis, Moving Average, Decomposition, Single Exponential Smoothing, and Double Exponential Smoothing. We use mean absolute present error (MAPE), mean absolute deviation (MAD), and mean squared deviation (MSD) to compare and select the least value. The result showed that Moving Average had the best performance. Then we used the Moving Average to verify and forecast over the next 12 months. However, it was found that the forecast values obtained were constant for the entire 12 months, so the moving average is unused for forecasting. The Moving Average has the least mean absolute present error (MAPE) at 0.2420. Therefore, we have used Decomposition which is a suitable performance in the second order of forecasting. It is forecast and has a trend value. Moreover, the Decomposition method has the least mean absolute present error (MAPE) at 0.1832.