A visualization framework for discovering prepaid mobile subscriber usage patterns

Abstract

Telecommunications operators are faced with a problem of knowing their prepaid subscribers behavior. And yet they have a challenge of reducing prepaid churn and maximizing the life time value of their subscribers. The prepaid subscriber is anonymous, the only way a prepaid subscriber gives information to the operator is through call records which are the recorded events on the use of the telecommunications network. The challenge is that the call details in their raw form do not provide any useful information. In addition, these details provide an overwhelming amount of data that is not easy to analyze. To assist the telecommunications operators, this study undertook to develop a visualization framework for discovering prepaid subscriber usage patterns. Exploratory approach was used to unravel subscriber usage patterns from call data records obtained from MTN Uganda. This was done using five visualization tools selected based on availability. These tools were used to explore the call records for patterns, while at the same time, a suitable visualization technique used by the tool to display the pattern was identified. Based on the findings, a visualization framework for discovering subscriber usage patterns is presented. The visualization techniques include; the 2-dimensional display techniques, for example time series, scatterplot suitable for identifying trends and outliers. Multi-dimensional display techniques such as parallel coordinates are suitable for viewing subscriber mobility, outliers, relationships, and common techniques like barchart, piechart are good for ranking patterns like high and low usage subscribers. The framework is evaluated using call data with known knowledge obtained from MTN Uganda. This framework is a starting point and as such it is open for further improvement.