Product recommendation system in the retail sector with data science and machine learning techniques

In this thesis, data science and machine learning techniques were examined to generate product recommendations in the retail sector. To achieve this, methods such as the Apriori algorithm, Recurrent Neural Networks (RNN), and Graph Based Analysis (GBA) were utilized. The performance of these methods was evaluated in detail on an example dataset from a gift shop and subsequently on a real-life retail dataset. While the Apriori algorithm offers a significant advantage in terms of creating simple and understandable rules in data sets, RNNs are effective in understanding long-term user behavior and making predictions on time series data. GBA stands out in subjects such as visualization of relationships, creation of communities and analysis of complex relationships. In studies conducted on the sample data set and real-life application, it was observed that the size and features of the data set were quite effective in deciding on the method. In data analysis and data visualizations performed using the Python programming language, the results were easy to read and accelerated in terms of adaptation to technology. The results of this study provide significant insights that contribute to the process of making personalized product recommendations in the retail sector. Choosing the appropriate one among the Apriori algorithm, RNN and GBA methods to increase sales rates and at the same time increase customer satisfaction is possible with a good understanding of data analysis and data features. Using technology-compatible tools such as Python in this data analysis accelerates the work.