Big data TV analytics for service and customer improvement

Abstract

Due to the rapid development of Internet Protocol Television (IPTV) technology, customers can now watch television on multiple devices with a vastly improved viewing experience. For a better understanding of customer experience and service performance, this has sparked a significant interest among service providers in the analysis of TV customer behaviour. Consequently, there is a high demand for TV analytics. However, research on IPTV customer behaviour is still insufficient, and there are obstacles to analysing the Big Data generated by IPTV systems.

This thesis presents a comprehensive study of data analytics techniques applied to the analysis of TV customer behaviour and IPTV system evaluation, based on IPTV datasets that occasionally reach the scale of Big Data. The study is comprised of three main sub-studies associated with various customer scenarios involving IPTV service interaction. Consequently, British Telecom (BT) provided three datasets, including gaze movement on the BT Set-Top Box (STB), clickstream on the STB, and labelled customers with clickstream in the three months prior to terminating their contract with BT.

A preliminary literature review on relevant topics was conducted in an effort to identify an appropriate data analytics technique and Big Data tool for this study. As a result, the widely employed Markov model served as the foundation for the methodologies used for separate studies. Subsequently, we developed three comprehensive methodologies utilising discrete-time Markov chains (DTMC), embedded DTMCs, and hidden Markov models (HMM), respectively, along with the application of the Big Data platform offered by BT. The studies have demonstrated the effectiveness of Markov models and generated insightful information regarding TV customer behaviour. In addition, the evaluation of the IPTV product and the investigation of customer engagement were completed. As a result, we discovered popular areas on the IPTV system page, common navigational paths on the IPTV system, and potential signs of frustration in TV customer behaviour.

Thesis embargoed until 31st May 2026

Date of Award	May 2024
Original language	English
Supervisor	Shuai Zhang (Supervisor), Sally McClean (Supervisor) & Gaye Lightbody (Supervisor)