Forecasting export value in the automobile industry

Abstract

The study aims to investigate the appropriate time series techniques for forecasting export values of cars and auto parts (Million Baht). The monthly data are gathered from the website of Ministry of Commerce from January 2010 to December 2017. The data are divided to two datasets. One dataset during January 2010 and December 2016 is consists of 84 observations which is used for build the forecasting model by using Winters' Additive Exponential Smoothing and Box- Jenkins. Another dataset between January 2017 and December 2017 are used for comparing forecast accuracy based on the lowest value of Mean Absolute Deviation (MAD) and Mean Absolute Percent Error (MAPE) and selecting the most appropriate forecasting model. The result shows that Box- Jenkins is the best method for this time series data.