A novel method for automatic discovery, annotation and interactive visualization of prominent clusters in mobile subscriber datasets

Customers are the most important aspect of any business and hence a solid customer segmentation strategy is a vital component in customer experience management (CEM). With declining revenues, increasing competition, regulatory pressures and price wars, communication service providers (CSPs) are increasingly focusing on CEM for subscriber retention and revenue enhancement. Grouping subscribers based on their behavior traits help CSPs to devise highly targeted marketing strategies and promotional schemes catering to preferences of individual segments, thereby improving the overall business performance and customer value. Clustering algorithms are widely used by CSPs for customer segmentation. Even though clustering algorithms attempt to identify natural groupings of subscribers based on their profile and service usage patterns, meaningfully visualizing and annotating these clusters to enable faster decisioning is a challenging problem, requiring a lot of manual intervention. In this paper, we present a novel scalable method for automatic discovery, annotation and interactive visualization of prominent segments in mobile subscriber datasets. We also extent this technique to segment migration analysis, allowing marketers to closely understand temporal behavior patterns of subscribers.