Analysis and Data Mining of Call Detail Records using Big Data Technology

Abstract: Call Detail Record (CDR) is a very valuable source of information in telecom industry; it opens new opportunities and option for telecom industry and maximize its revenues as well as it helps the community to raise its standard of living in different ways. However, I need to analyse Call Detail Record in order to extract its big value which helps to find new business opportunities. Real time streaming data processing is became new trends in Call Detail Record processing. It helps to analyse Call Detail Record in real time and helps in finding real time location of any customer and also behaviour of network in real time. But these Call Detail Records has a huge volume, variety of data and different data rate, while current telecom systems are designed without considering these issues in mind. Call Detail Records can be seen as Big Data source, and hence, it is applicable to use Big Data technologies (for storage, processing and analysis) such as Hadoop in Call Detail Record analytics. There are considerable research efforts to address the Call Detail Records analysis challenges face in telecom industry. This project presents the use of Big Data technology in Call Detail Records analysis by giving some Call Detail Record analytics based application examples, highlighting their architecture, the utilized Big Data and Data Mining tools and techniques, and the Call Detail Record use case scenarios. In this project I am processing Call details record and calculating traffic in Erlang and using this I am creating cluster of base stations and checking its performance for different new pricing and bundling can be seen and management decision can be made.

Keywords: Convergent billing, K-means clustering algorithm, machine learning, normalisation, bundling, Hadoop.