The supervised CDR dataset used in this approach was acquired through one of the leading telecommunication service providers, Dialog, in Sri Lanka. It was collected from Dialog Axiata over 3 months, from April 2017 to August 2017. More timely expanded semi supervised data set was provided by Dialog during the research period. A survey was conducted on a sample of customers to identify whether they are using an additional mobile subscription or not in order to prepare the supervised dataset. The data was provided as separate data tables containing customer data, device data, CDR related to calls, SMS and GPRS along with labels for multi-SIM and non multi-SIM users. Unique ID of each customer was used to integrate these data into a single dataset. The collected data was about 2375 users, out of which 2194 were prepaid users and the remaining 181 were postpaid users.
Associated Publication: -
Paper Title: Categorical Classification Approach for Identifying Multi-SIM Users from Call Detail Records
Published in: 2019 National Information Technology Conference (NITC)
Date of Conference: 8-10 Oct. 2019
DOI: 10.1109/NITC48475.2019.9114444
Further, a paper on "Categorical Classification Approach for Identifying Multi-SIM Users from CDR" was submitted in ICDM – International Conference on Data Mining.
Citation: -
C. Soysa, S. Karunathilaka, A. Matharaarachchi, H. Rodrigo and U. Thayasivam, "Categorical Classification Approach for Identifying Multi-SIM Users from Call Detail Records," 2019 National Information Technology Conference (NITC), 2019, pp. 66-71, doi: 10.1109/NITC48475.2019.9114444.
