Modeling Factors of Product Classification Proximity in Data Mining Applications

Abstract

In today's nano-technology era, as device sizes shrink, the volume of data generated has grown exponentially. This increase manifests in both the number and size of files. Real-world data corroborates this trend. Given the massive scale of data, efficient methods are required to extract and process information to meet specific needs and preferences. One promising approach is data mining, particularly in analyzing the relationships between frequently purchased products. For instance, in a convenience store, if a customer selects product Z1, what is the likelihood that they will also purchase product Z2? How does this pattern extend to subsequent purchases? By understanding these associations, retailers can optimize product placement, enhancing customer convenience and increasing sales. Furthermore, retailers can streamline procurement processes by focusing on products with strong association patterns, thereby reducing costs. This research employs data mining techniques, specifically the Apriori algorithm, to investigate product associations in a convenience store using data collected between May 1 and September 30, 2023. The CRISP-DM methodology guided the research. The findings reveal significant associations between products purchased by customers, as evidenced by the highest frequent itemsets and association rules. Additionally, the data underscores the need to classify consumer goods into food and beverage categories. The analysis indicates that the purchase of food items does not necessarily lead to the purchase of beverages, and vice versa. Consequently, it can be concluded that a customer's purchasing decisions are primarily driven by their immediate needs or preferences rather than established product associations or habitual behavior.