Identification of Distinct Usage Patterns and Prediction of Customer Behavior

Sharam Dadashnia, Tim Niesen, Philip Hake, Peter Fettke, Nijat Mehdiyev, Joerg Evermann

In: Sixth International Business Process Intelligence Challenge (BPIC’16). Business Process Intelligence Challenge (BPIC-2016) befindet sich BPI / BPM 2016 September 19-19 Rio de Janeiro Brazil Springer 2016.


The given BPI Challenge 2016 provides a case study based on a reallife event log. In this report, we analyze usage data from IT systems of the Dutch employee insurance agency (UWV). The data comprises information about customer demographics, click data describing their behavior when using the agency's website, and data from customer service systems. We identify distinct usage patterns, report on the change of those patterns over time as well as on cause-effect analyses explaining when customers deviate from standard procedures. Moreover, we present a prediction approach based on deep learning algorithms that helps to determine future events for running customer sessions. As a result, some recommendations for the UWV are derived in order to increase the user experience and decrease expensive communication-channel transitions.

bpi_challenge_report_submission.pdf (pdf, 1 MB)

Deutsches Forschungszentrum für Künstliche Intelligenz
German Research Center for Artificial Intelligence