Customer: A novel customer churn prediction method based on mobile application usage

Abstract

Technological innovation and fierce dispute to conquer the mobile device market make manufacturers and telecommunication companies increase their attention to the loyalty of their clients. In this sense, customer churn prediction is particularly important, and recently received more attention from aforementioned companies, mainly those that merchandise mobile devices. Besides helping users in their daily activities, these devices provide to manufacturers a large amount of data, as application usage and battery consumption. It is important for companies to know how customers use their applications. Automated tools are needed to collect this large amount of data and help companies to understand customer's behavior throughout time, seeking to prevent customer abandonment. Based on this, this work proposes an innovative and effective method to manage and predict customer churn based on mobile application usage. In order to investigate the distribution of data throughout time, the proposed method processes application usage data, based on techniques of data stream clustering (where customers are grouped by their behavior) and based on novelty detection. Our tests show an accuracy rate of 87%, a satisfactory performance in this context.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 6 citations exist upstream.