Analysis of Call Detail Record and mining the user behaviour for fast decision making using Big Data Technology

Abstract

Call Detail Record (CDR) is Telecom companys
most valuable source of information. It gives information about
customers behaviour and usage details which is useful in billing.
And this information is also used in Telecommunication industry
for it’s fundamental processes (i.e. billing,charging, settlement,
network efficiency check, fraud detection and avoidance, revenue
assurance, churn detection and prediction , value added services,
business intelligence, etc.). Call Detail Record is a record contains
detailed information about a users event by event transaction,
such as call start time, call end time, call duration in seconds,
called parties identification, cell ID records, requested websites,
type of data i.e. if it is calling or internet usage. Moreover,
Call Detail Record may help to improve many existing business
processes and services such as business intelligence, marketing,
transportation’s and networking etc.
Telecom industry is considered as the most competitive and
healthy profit industries. As advancement in the distributed
computing, each and every company tries to gain quick knowledge from available data-set. BY knowing more about customers
and their customer behaviours helps them to make appropriate
marketing and network planning and management decisions.
Quick decision helps them to be more competitive and helps
in operational cost reduction. This call detail record data-set is
very huge in size.
In this paper we proposed an application for clustering base
station based on traffic that helps to analyze segment of data,
helps in grouping of customer according to their daily usage of
communication network (such as Erlang based on call duration)
and services supported by a telecom operator. In this paper we
are presenting tool that uses Scala language and Sparks API for
processing this large data-set. This behaviour can be shown as
line graph for managerial people for decision making.