Partitional clustering based on PCA method for segmentation of products

This paper focuses on grouping e-commerce data for product segmentation with dimension reduction using Principal Component Analysis (PCA) on the data to transform the original data to the top principal components' feature space. K-means based on principal components is used for forming the clusters. Then, we conducted the performance evaluation using metrics to calculate the goodness of the clustering technique. The segments derived from PCA in conjunction with k-means are able to provide interesting insights for data-driven decision making in practice. The results indicate three segments of products: The least sold products with an acceptable rating; the most expensive and highest rated products; the more economical and best-selling products.