A neuro-fuzzy based forecasting approach for rush order control applications

This paper proposes an adaptive neuro-fuzzy inference system (ANFIS) and a KERNEL System to solve the problem of predicting rush orders for regulating the capacity reservation mechanism in advance. The adopted approaches generalize the association rules among rush orders, as well as to forecast product items, quantities and the occasion of the contingent rush orders via learning from the sales data of an actual electronic manufacturing firm. Especially, we compare results with the traditional regression analysis and obtain preferable forecasts. In sum the overall forecasting correctness is 83% by ANFIS which is superior to regression manner with 63%. Preliminary results on the application of the proposed methods are also reported. It is expected to offer managers to refer to arrange the reserved capacity and to construct a robust schedule in anticipation.