Hierarchical Naïve Method for Short Term Vehicle Sales Forecasting
Vehicle demand forecasting is a classic problem in the automotive domain, traditionally used for inventory optimization in supply chain management. Implementation of centralized dealer management systems by OEMs has resulted in generation of granular level retail data paving the way for advanced data based applications. This paper implements demand forecasting for applications wherein short-
term extrapolation is required, such as forecasting mid-month the retail sales at the end of a given month. In these circumstances, our findings show naïve methods to perform better than sophisticated statistical forecasting models. Furthermore, an innovative method using hierarchical modelling techniques has been introduced which we have named ‘Naïve Hierarchical Forecasting Model’. We foresee its application in – forecasting the end of month sales for timely intervention towards achieving monthly targets; and analyzing a dealer’s performance with respect to the set targets over a period in improving the target setting practice itself. The algorithm is packaged in a data application created using QlikView. We foresee its application in – timely intervention to achieve the overall monthly target of dealers; analyzing a dealer’s performance with respect to the set targets over a period in improving the target setting practice itself, and for analyzing performance of products by dealers in different states.