Modeling Salesperson Performance Based On Sales Data Clustering

The case study in this research is a company in Indonesia. Currently, the company needs to assess the performance of salesperson, but it does not have yet criteria for classifying the salesperson's performance. The research purpose was developing salesperson performance model based on sales data clustering. The research is adopted CRISP-DM framework. As the result, the proposed model is connected with the prior sales order database. The model grab multidimensional features and classification data label from the database. The multidimensional features was formulated through Kohonen SOM clustering evaluation result with the value of quantization error was 0.95 and topographic error value was 0.13. The best multidimensional features are product, transaction, and monthly price. It can be concluded that the dimensional combination is considered capable of representing salesperson performance. Therefore, it can be used as base criteria or features in classification. The contribution of the research is built a new model of salesperson performance for the company.