Determining product categories sales relationship with association rules and cluster analysis: an application in retailing sector

Many enterprises have lots of data about their customers. Processing these data with various analysis techniques to make meaningful information for enterprises will make enterprises to take beneficial decisions in the competitive world. Data mining determine hidden informations and patterns in a data set with the help of mathematical and statistical techniques to help enterprises to make choices.In this research it is aimed to determine the sales relationship between product categories and product classes. To achieve this goal, data mining methods, like association rule analysis and hierarchical cluster analysis, applied to one year sales database constructed by a retailer. In this research 1.096.125 sales slips were analyzed. In association rules analysis, the biggest rule support (%18.69) was found between beverages and sweet categories, the highest confidence value (%69.46) was found between private label and sweet categories. In cluster analysis the strongest relationship was found between macaroni-legumes (0,403), olives-cheese (0,354), tea-sugar (0,337), pastry ingredients-ready-made sweet (0,308) classes. Clustering stage was completed in 82 steps.Results of this research can be used in promotion campaigns by retailers. Product categories and classes, which retailers use in category management, can be formed efficiently by examining cluster analysis results. Patterns, which were founded by this research, can be used to design activities for indecisive customers to persuade them.