Databases Reference
In-Depth Information
The hardest part of collecting and analyzing network data is that it's all semi-structured
and streaming in nature. Largely, telecommunication companies have built systems to
detect critical outages and bottlenecks and raise alerts, but these systems were rarely
designed to analyze network performance over a long period. If you have the ability to
understand how and where service issues are trending and how that is affecting your
most profitable customers, then you can think of ways to improve.
By combining dropped calls data and latency for video-based services with
subscriber's dynamic and static information, you can identify cell towers that are
performing poorly and impairing the service experience. This approach can enable
operators to analyze and get better insight to network performance and quality of service
from a customer's perspective and help them to take proactive measures to answer
questions like the following:
•
Which regions in my network had the most dropped calls in the
past hour, day, week, and which of my customers were most
affected? Are these customers profitable? Are they likely to churn?
•
Is this a one-off scenario, or it is actually a trend? How can
I prioritize where I should invest new capacity in my network,
based on customer revenue and profitability?
•
Which of the outages were due to handset problems, wireless
coverage problems, or switch problems?
•
Is my network performance breaching SLAs that have been
agreed upon with certain customer segments? How can
I prioritize the traffic of those customers in order to avoid
SLA breach?
By combining call detail records (CDR) data, cell-site data, calling-circle data, and
social network data, you can identify communities and social leaders. This approach can
enable operators to quickly determine who are the leaders (“ARPU inducers”) and who
are the followers (“churn shielders”). These insights can be effectively utilized to develop
pricing at an individual level. In addition, the social network monitoring in real time can
help in managing “word of mouth” (negative and positive publicity) events based on early
detection of social chatters.
Big Data Analytics for Banking
Customer focus is increasingly becoming important for many financial organizations.
While customer analytics is not a new concept for banks, some of the best known
customer analytics use cases have come from the banking industry: fraud detection, risk
analytics, credit scoring, and anti-money laundering are prime examples. However, in the
Internet era, the growth of data is posing serious challenges to these customer analytics
applications. Banks need new technologies to handle the unprecedented volume, variety,
and velocity of information.
Figure
3-5
illustrates a typical banking applications landscape overlaid with the type
of data sources (structured or unstructured) and big data characteristics.
