<![CDATA[UD Dwi Mulya Plastik is a company that produces plastic ore based house appliances. To fulfill customer’s demand, it requires sales forecasting so it can reduce the risk of over production or under production. The sales is represented as time-series data. In this work, we utilized an Artificial Neural Network method that commonly called Improved Elman method to do forecasting of the time-series data. Based on our examination, it is shown that the smallest value of averaged MSE (0,29) as well as MAPE (12,69%) for ‘timba cor’ can be achieved by using learning rate of 0,50 and the number of input data 12. On the other hand, the smallest value of averaged MSE (0,011) as well as MAPE (15,23%) for ‘waskom’ can be achieved by using learning rate of 0,30 and the number of input data 12. It is commonly understood that the maximum value of MAPE to be categorized as good forecasting is 20%, hence, it is concluded that the Improved Elman method in this study is considered valid.]]>
Copyrights © 2017
