Improving customer segmentation via classification of key accounts as outliers

Abstract Customer segmentation and key account management are important use cases for clustering algorithms. Here, a data set of Portuguese wholesaler food household supplies is used as an exemplary application. To increase the quality analysis, two-stage approach proposed. First, accounts filtered by density-based outlier detection. Second, Gaussian Mixture Model (GMM) applied to cluster smaller customers. This aligned with business implications outstanding very differently behaving customers well core idea ABC analysis. Also, exclusion corresponds definition outliers results different underlying mechanism. Using this shows better compared using one-stage applying only GMM. Therefore, it concluded that detection followed GMM beneficial customer within B2B applications.