A predictive model for recurrent consumption behavior: An application on phone calls

• Published in: Knowledge-based systems 2014-07, Vol.64, p.32-43
• Main Authors: Haddad, Mohamed Ramzi, Baazaoui, Hajer, Ziou, Djemel, Ben Ghezala, Henda
• Format: Article
• Language: English
• Subjects: Applications programs; Business; Consumer behavior; Consumers; Exponential moving average; Inhomogeneous poisson processes; Knowledge base; Mathematical models; Mobile communication systems; Phone calls; Phone calls prediction; Statistical modeling; Weight reduction
• ISSN: 0950-7051; 1872-7409
• DOI: 10.1016/j.knosys.2014.03.018

Nowadays, companies use different datamining and prediction technologies in order to better forecast demands, consumers interests and business requirements. Anticipating the future helps businesses being proactive, managing resources and making intelligent decisions and investments. In this article, we propose a prediction model for recurrent consumption behaviors based on inhomogeneous Poisson processes aiming at predicting users’ future incoming and outgoing phone calls. The proposed model is lightweight in terms of processing power and storage requirement, capable of detecting users’ recurrent phone calls and self-adapting to their changing behaviors and trends. The calls prediction model was implemented as a mobile application and evaluated in real world conditions. During 12months, different configurations of our model were evaluated on a set of 7645 phone users in order to better tune it and measure its predictions quality.