T-HT Big Data pilot project

Telecommunication sector

Working on this project was challenging and instructive, because dealing with three BigData practical business casees we had the opportunity to apply the latest BigData technologies such as sentiment analysis, analytics streams for processing large amounts of data (Internet of Things) using Hadoop as BigData Storage. In addition, using the latest technology and architecture to practical business cases it was very motivating and inspiring experience for us.

Luka StepinacProject manager, CROZ d.o.o.

Published 2015-12-01 16:15:58

Main activity of Hrvatski Telekom d.d. (Croatian Telecom Inc.) is the provision of telecommunications services, design and construction of communications networks on the Croatian territory. The Group also provides fixed telephone line service (access and traffic through fixed telephone lines and additional fixed network), Internet services including IPTV, data transmission services (lease of lines, Metro-Ethernet, IP / MPLS, ATM) and operates GSM and UMTS mobile telephone networks.

Highly developed technological infrastructure facilitating further development of services is a guarantee of growth and potential. This is the reason why Hrvatski Telekom d.d. chose to implement a Big Data platform based on IBM technologies, with CROZ experts.

A Big Data Platform ready for Business Scenarios

Enterprises today must store and analyze more data from more sources than ever before. One of the goals of this project was to prove a thesis that Hadoop, as an underlying technology, provides a flexible, low-cost, proven platform over traditional data containers for long-term growth.

After successful installation and configuration of hardware and software assets, several data sources were chosen for project. For data-at-rest, existing CRM and Data Warehouse were used, as well as social media data collected using the CROZ Sentimenter solution. For Data-in-motion, streaming process was established for collection of data in real time from mobile stations / cells generating CDRs (Call Data Records); approximately 7 billion CDRs were analyzed. IBM SPSS was used to explore hidden patterns in data and create different customer segments, revealing great insights for focusing of campaign targets. Also, interesting anomalous behavior was detected in several situations, such as extreme upload from a specific handset model, high traffic via Viber request from one handset, malware prediction from HTTP requests and so on.

IBM SPSS Modeler rezultati analize

Photo: IBM SPSS Modeler analysis results

Customer Segmentation by Data Volume – Anomalous Segments

Using the above platform and data stored in IBM BigInsights, which is an IBM’s enterprise distribution of Hadoop enriched with GUI & admin tools for better efficiency, following benefits were realized during the project:

  • It was the first project which made successful connection to xDR distribution (Gn interface) from non NT environment
  • Completely new types of data were made available for analytics and correlation with existing sources
  • Business scenarios for leveraging data were created and prioritized, e.g.:
    • Social Media Sentiment Analysis,
    • Location Based Campaigns,
    • Competitor Offer Monitoring,
    • Inbound Roamers Tracking,
    • Email classifications aso.
  • Confirmed technology & architecture fit, concept of platform vs. concept of tool

IBM BigInsights korisničko sučelje

Photo: IBM BigInsights user interface

With this innovative project, Hrvatski Telekom d.d. confirmed its strategical direction as the leader in the field of innovation. Considering their large network of current and future users, we can expect a direct influence on the advancement of the customer experience

Other projects

All projects