Change Point Detection Based on Call Detail Records
In this paper, the authors propose a method for combining wavelet denoising and sequential approach for detecting change points on mobile phone based on detailed call records. The Minmax method is used to estimate the thresholds of frequency and call duration for denoising. This paper is useful to enhance homeland security, detecting unwanted calls (e.g., spam) and commercial purposes. For validation of their results, they randomly choose actual call logs of 20 users from 100 users collected at MIT by the Reality Mining Project group for a period of 8 months.