A knowledge discovery based approach to long-term forecasting of demand for electronic spare parts

This paper deals with modeling and predicting purchase life-cycles of electronic spare parts that are supplied to the repair vendors by companies which provide the so-called spare parts logistics service. We introduce a soft computational method that can be used to discover the typical purchase life-cycles of end-of-life spare parts that belong to the same commodity class. In our approach, the discovered knowledge is embodied by typical demand models which can be utilized to forecast the demand for active spare parts, that is, for components for which there are current demands. We apply a fuzzy similarity based approach to generate the forecast from the typical demand models. The introduced forecasting method is advantageous in long-term prediction, it can be especially useful in supporting purchase planning decisions in the ramp-up and declining phases of purchase life-cycles. The application of our method is demonstrated through real-life examples.