Large Scale Data Processing • Machine Learning Model Development & Implementation • Real-time Analytics • Customised Advertising Solutions
The Customer:
T-Mobile Hungary, part of the global telecommunications group, is a leading provider of mobile communications services in Hungary. Always at the forefront of innovation, T-Mobile Hungary continually seeks opportunities to enhance its offerings and create new revenue streams.
Project Summary:
To tap into the potential of location-specific advertising, T-Mobile Hungary envisioned a system that could predict the movements of its subscribers. We collaborated to create a large-scale Machine Learning (ML) solution that would not only process real-time data from T-Mobile's Intelligent Network but also forecast the likelihood of a subscriber passing a particular location.
Services:
• Large Scale Data Processing
• Machine Learning Model Development & Implementation
• Real-time Analytics
• Customized Advertising Solutions
Overview:
The core of the project revolved around harnessing the power of eXtended Detail Records (XDRs) - encompassing call, message, and internet details - and processing them in real time. By analyzing this data, the goal was to predict the probable paths subscribers might take, enabling businesses to push timely and relevant advertisements.
Challenges:
The vast volume of real-time data presented significant challenges. Processing and analyzing millions of XDRs daily required a robust system. Additionally, the predictions had to be highly accurate to ensure the relevance of the advertisements and to avoid any negative user experience.
Solution:
We utilized the Green Plum Database technology, known for its capacity to manage large datasets, and combined it with custom Java code to handle the data influx. The heart of the solution was the ML model, designed to analyze patterns and behaviors from the XDRs. By understanding a user's routine and frequent locations, the model could predict with high probability the future paths a subscriber might take.
For instance, a car dealership could strategically place ads for a user who was predicted to pass by their showroom, ensuring visibility before, during, and after the potential visit.
Outcome:
The platform was a resounding success. By leveraging real-time data and accurate predictions, T-Mobile Hungary could offer businesses a unique advertising opportunity. The resulting advertising inventory not only provided businesses with a novel way to reach potential customers but also generated a significant new revenue stream for T-Mobile Hungary.
