Improving customer segmentation via classification of key accounts as outliers
Open Access
- 30 September 2022
- journal article
- research article
- Published by Springer Nature in Journal of Marketing Analytics
- Vol. 11 (4) , 747-760
Abstract
Customer segmentation and key account management are important use cases for clustering algorithms. Here, a data set of a Portuguese wholesaler for food and household supplies is used as an exemplary application. To increase the quality of the analysis, a two-stage approach is proposed. First, key accounts are filtered by a density-based outlier detection. Second, a Gaussian Mixture Model (GMM) is applied to cluster smaller customers. This two-stage approach is aligned with the business implications of key accounts as outstanding and very differently behaving customers as well as with the core idea of an ABC analysis. Also, the exclusion of key accounts corresponds to the definition of outliers as the results of a different underlying mechanism. Using this two-stage approach shows better clustering results compared to using a one-stage approach applying only a GMM. Therefore, it is concluded that density-based detection of key accounts followed by a clustering using a GMM is beneficial for customer segmentation within B2B applications.Keywords
Funding Information
- Karlsruher Institut für Technologie (KIT)
This publication has 27 references indexed in Scilit:
- Introduction to Data Science2017
- Computed ABC Analysis for Rational Selection of Most Informative Variables in Multivariate DataPLOS ONE, 2015
- Enhancing the selection of a model-based clustering with external categorical variablesAdvances in Data Analysis and Classification, 2014
- A neuro-fuzzy two-stage clustering approach to customer segmentationJournal of Marketing Analytics, 2013
- ‘All models are wrong...’: an introduction to model uncertaintyStatistica Neerlandica, 2012
- Anomaly detectionACM Computing Surveys, 2009
- The Elements of Statistical Learning2009
- Assessing a mixture model for clustering with the integrated completed likelihoodIEEE Transactions on Pattern Analysis and Machine Intelligence, 2000
- A decomposition of multidimensional point sets with applications to k -nearest-neighbors and n -body potential fieldsJournal of the ACM, 1995
- Silhouettes: A graphical aid to the interpretation and validation of cluster analysisJournal of Computational and Applied Mathematics, 1987