Abstract:
This research was aimed to study the sales forecasting models for packaging coating type P1 and P3. The monthly sales volume of packaging coating type P1 and P3 collected from January, 2011 to December, 2018 Were divided into 2 Sets. The first set of data from January, 2011 to December, 2017 were used for constructing and selection the forecasting models by Triple Exponential Smoothing method, Holt-winters Exponential Smoothing method, Box-Jenkins method and Artificial Neural Networks and employed Root Mean Square Error (RMSE) as the criteria for selection the models. The second set of data from January, 2018 to December, 2018 Were used for comparing the performance of the forecasting models and employed Mean Absolute Percentage Error (MAPE) to show percentage of error. The forecasting models from Box-Jenkins method are the most appropriate models for Packaging Coating of both P1 and P3 and has the highest forecasting accuracy which the RMSE are 2,701. 6369 and 11,017.4307 and the MAPE are 18.3568% and 13.0210% respectively. When comparing the forecasting results between Box-Jenkins method and the current company method, it is found that Box-Jenkins method gives lower inventory. By employing Box-Jenkins method to forecast Packaging Coating sale volume, the company can save 13,149 and 132,849 kilograms per year for P1 and P3 inventories respectively.