Aplikace pokročilé datové analytiky v prostředí e–commerce společnosti

This master thesis deals with advanced data analytics in e-commerce business. Nowadays, e-commerce companies possess large amount of data from a variety of data sources with the potential to leverage that data for their competitive advantage. Thus, the objective of this paper is to build a recommender system model for cross-selling using item-based collaborative filtering approach utilizing data from a specific e-commerce company that could be deployed in production afterwards. In the first stage, cluster analysis of customer RFM segmentation variables is performed using the k-means method with k = 5. Subsequently, the item-based collaborative filtering RS is applied on each of the obtained clusters separately. The recom- mendations are generated on the product category level, with each category being mapped to its best-selling product as the final output of the recommender system. The number of top-N recommendations can be chosen as a parameter. The evaluation yields a result that 45.4 % of customers of validation set being correctly recommended at least 1 category. Considering the data sparsity of the user-item matrix, further research and model development might be conducted to optimize the mapping rule of products to the recommended categories alongside best-selling products to minimize the observed long tail effect.