Explainable Customer Segmentation Using K-means Clustering

Abstract

Explainable AI has gained popularity in recent years, but the application of it in unsupervised learning is still a few. In this research, explainability was integrated with clustering, an unsupervised method. Customer segmentation is one of the most important aspects in the competitive business world. The most common approach for customer segmentation is clustering, however, assignments of the clusters often can be hard to interpret. To make the cluster assignments more interpretable, a decision tree based explainability was implemented for customer segmentation in this research for small and large datasets. Using the Elbow Method and Silhouette Score, an optimal number of clusters were found, then ExKMC algorithm was implemented for both datasets.