Data analytics in industrial applications

Information Technology Distributors need to develop methods for understanding their dealers purchasing behavior regarding to product groups in order to prepare their business plan correctly. So that correct customer segmentation is seriously important for those companies. This study provides customer management framework that is provided in order to increase IT distributors sales. Basicly, this framework includes 3 phases which are data preparation, effective clustering and prediction. Real application of proposed method is provided for one of the biggest IT distibutor company which is in Istanbul. Firstly, K-means, expectation maximization algorithm (EM) and hierarchical clustering methods ,which are proper for the data set, are applied IT dealers sales data. As a result of clustering process, it is observed that hierarchical clustering provides better performance compared to others. This study also provides decision support tool that suggests right product groups in order to increase IT distributors sales with classification method. Proposed method uses clustering results as preprocess of prediction process. Sales potential of dealers is determined regarding to pruduct groups with classification algorithms that are Naive Bayes, Support Vector Classification and C4.5 algorithms. Finally the business plans that are created via proposed framework have potential to yield results.