Visual discovery and analysis of voice traffic between telecommunication companies in Uganda

Abstract

Telecommunication companies receive a vast amount of data from various sources. Data is electronically captured in OLTP systems and is further processed to data marts and data warehouses. This data is usually appreciated from a quantitative view which limits the quick and easy identification of irregularities, patterns, trends and relationships from within the data. The facts are identified by aggregations. The main objective of the study was to visually explore data on traffic between originating and terminating routes between telecom partners in Uganda. From the investigations the results showed that the originating calls from UTL and Celtel were a mere fraction of those from MTN. The calling patterns from UTL and Celtel were varyingly irregular with large drops on average over the weekend despite low tariffs. The low tariffs did not seem to motivate the subscribers. A trend showed that the subscribers call highest between 2000 and 2100 hrs and during the office hours the trend showed a possible use of cheaper alternatives to communicate such as office phones (paid for by the company), web SMS (like www.trueafrican.com), emails, meetings to mention but a few. Visual exploration of data enables an end user to easily understand inhomogeneous and noisy data. This is true to the dictum that a picture is worth a thousand words.