Abstract:

The extracting association rules of inter-user-product relations used by companies in decision-making processes have been popular for some time, especially for market basket analysis. In this study it is aimed to discover association rules from original online store transaction of a Turkish retail company, in order to help administrator and decision maker also Customer Relationship Management department to initiate campaigns. The main objective is to find out which product item sets are purchased together. In order to better compare the results the data are analyzed with and without clustering according to the range of ages and gender. Data mining Association analysis methods such as Apriori Algorithm, FP-Growth (Frequent Pattern) then applied which are used to extract association rules. Moreover some of the collaborative filtering metrics namely Jaccard, Pearson, and Cosine function are used to understand the association between products to obtain a recommendation system. The proposed recommendation methods successfully recommended the associated product for the obtained original dataset as high as 65% accuracy. Obtained association rules are shared with the marketing department to initiate and direct forthcoming marketing campaigns.

Keywords: