Modelling and Forecasting for Automotive Parts Demand of Foreign Markets on Thailand

Abstract

Thailand’s export of automotive parts to foreign markets increases in value greatly each year. Within this group, the wheels, including parts and accessories (WPA) exported to foreign markets are of the greatest value to manufacturers who need to be aware of the demand in advance. The objective of this research is to find an accurate forecasting model for predicting advanced demand for the WPA and calculate the optimal quantity for export by using a linear programming (LP) model in order to benefit from the maximum profit. The methodology is selection and analysis in a performance of time series forecasting models (naïve, moving average, single exponential smoothing, and exponential smoothing with trend) and an artificial neural networks (ANNs) model in forecasting the WPA exported from Thailand. The countries with the five top highest demands for the number of WPA are: Japan, China, South Korea, Germany, and Indonesia. The demand data of five countries is used for analysis and the data was collected during the period of 1997 to 2008. The mean absolute percentage error (MAPE) is used as a measure of forecasting accuracy. The results reveal that an ANNs model outperforms the other models for Japan, Germany, and Indonesia, whereas an exponential smoothing with trend is close to the actual demand for China. There are, however, no accurate forecasting models for South Korea because the five models give an error greater than 50%. The accurate forecasting model for each country is used to forecast the advanced demand. After that the forecasting results are used for finding the optimal quantity for exporting to the five countries by using the LP model in order to receive maximum profit. The results of this research can be used for effective production planning or investment by informing manufacturers.