Analysis of e-Commerce platform sales data using data mining techniques

In this master's thesis, sales data from a large-scale e-commerce platform were analyzed using data mining techniques, with the objective of conducting a multidimensional examination of customer behavior. The study specifically focuses on two fundamental data mining approaches: Market Basket Analysis (MBA) and the extraction of association rules between products. Within the scope of the research, the Apriori, FP-Growth, and H-Mine algorithms were implemented both in a single-core Python environment (mlxtend) and on the highly scalable Apache Spark framework, utilizing varying levels of parallel processing power. This dual implementation enabled a comparative evaluation of the performance and efficiency of traditional methods versus distributed/big data processing platforms. Analyses were conducted on datasets ranging in size from 100 rows to 10 million rows, and the algorithms were comprehensively assessed in terms of execution time, rule generation capacity, scalability, and computational resource utilization. The results revealed statistically significant association rules that effectively reflect customer purchasing tendencies. High-support and high-confidence product combinations were identified, which can be directly leveraged in personalized marketing strategies, cross-selling applications, and the enhancement of recommendation systems. Additionally, the advantages and limitations of different association rule mining algorithms were comparatively analyzed based on computational cost, memory consumption, and rule quality in large-scale data environments. This study serves as a comprehensive example that demonstrates the high potential of data mining methodologies in the e-commerce domain. From an academic perspective, it contributes to the literature on association rule mining both methodologically and in terms of applied research. From an industry perspective, it offers scalable and applicable solution strategies capable of providing a competitive advantage for enterprises in big data processing, customer behavior modeling, and data-driven marketing strategies.