Application of K-means Clustering Algorithm in Consumer Shopping Behavior Segmentation in E-commerce

Companies are forced by fierce rivalry to be the best at satisfying customer requirements in order to keep clients from moving to rivals. Therefore, an algorithm—such as the K-means Clustering algorithm—is required to segment customer buying behavior so that businesses may more accurately satisfy demands. Using a simulated dataset from Kaggle that contains nine variables with information on consumer purchasing behavior, this study attempts to apply K-means Clustering to segment customer shopping behavior in e-commerce and assess its efficacy. Pre-processing, normalization, and the selection of three important numerical features—AvgTotalSpend, AvgSpendPerTrans, and LoyaltyScore—are all part of the analytic process. The elbow technique and silhouette score are used to determine the ideal number of clusters. The density between clusters is also evaluated using the Dunn index. Three separate consumer clusters are identified by the segmentation results: Cluster 0 has very low values for AvgTotalSpend, AvgSpendPerTrans, and LoyaltyScore; Cluster 1 has reasonably high values for all three attributes; and Cluster 2 has low LoyaltyScore but high AvgTotalSpend and AvgSpendPerTrans. According to these results, customers in Cluster 1 are more likely to make repeat purchases in the future, which offers insightful information for focused marketing campaigns.