The wide product offering made available by supermarkets and online stores provides variety and choice to its customers. This allows stores to target a wider variety of customers, increasing its overall revenue. On the other hand, the wide offer makes it increasingly difficult for customers to make the right choice according to their personal taste and requirements. Analysing the whole offer to choose the right product is not always practicable for costumers due to time and availability restrictions. It would therefore be useful to be able to provide personalized recommendations to customers, helping them to find the most suitable and interesting products for them. An interesting opportunity arises from the information possessed by stores about its customers and products. Information such as the purchase history of customers can provide an important insight into customers’ buying habits and can be used to segment customer profiles.
This project aimed at creating a system that analyses purchase history data to automatically create personalized product recommendations to customers. Data mining techniques were applied to analyse the data in order to create recommendations according to different criteria. The project had the cooperation of Sonae Modelo Continente, who supplied real purchase history data for the development of the system, making it more realistic.
Author: Luís Moreira
Type: MSc thesis
Partner: Faculdade de Ciências da Universidade do Porto