Summary: | 碩士 === 明新科技大學 === 管理研究所碩士在職專班 === 106 === This study proposes Grey System models (GM) for the top 1 material ( Gold wire) usage forecasting semiconductor packaging and testing. The MAPE, MSE and Theil’U inequality coefficient were being used to evaluate the predictive errors of the models proposed. The time series data is obtained from SAP System. The empirical results show that Data during 2013 to 2015, NGBM model has a better predictive performance than GM(1,1) model and Arima model. In this research, we confirmed that the prediction error rate of the GM is less than 15%, therefore, that it can be applicable to medium-term time service data forecasting.
|