Applied Data Science Research

How Profinit combines scientific research with applications in finance

by Barbora Janulíková, Business & Data Science Development Manager

If you have been following Profinit for a while, you know that academic research is deeply encoded in our company‘s DNA. We only pick new colleagues with the best education in mathematical and technical sciences. Moreover, our consultants teach Software Engineering or Big Data technologies at top universities. This cooperation is mutually beneficial: applied research promotes innovation in the business world, and practical applications are a great source of exciting research topics.

In addition to these efforts, we are enormously proud of our continuous project with the Faculty of Mathematics and Physics of the Charles University, Prague. Profinit’s Data Science team is currently developing a new system for advanced Big Data Analytics. The rich experience we gained from working on Big Data projects with banks, insurances and telcos serves us well as a knowledge basis.

Imagine the following situation: Clients contact a bank asking for a specific product or service (e.g., travel insurance, credit card, etc.). If the bank had free access to social networks, it would be fairly obvious that these clients have something in common (e.g., they work for the same company, are part of the same social group, partake in same hobbies, etc.). Subsequently, the bank could easily offer the same products or services to other clients based on their similarity in these areas.

The issue is of course that the banks do not have access to their clients´ social networks. And this is precisely the problem we address in our research project. By applying the most recent techniques from statistics and probability modelling to a large pool of transactional data, we have been able to build a so-called ´pseudo-social network´.  As a result, we can extrapolate the client´s social network from his banking transactions. Our experience shows that such a reconstructed context (or if you will, the clients´ similarity environment) brings extra value in the areas of, e.g., marketing or risk and predicts many business indicators with high precision.

In real life, banks cannot access the desired information about a certain client. However, based on individual context and attributes of other people, who are similar to any given client in their behaviour, our model can predict his/her credit score, probability of default, and much more.

Profinit has been supplying the banking industry with its Data Science expertise and verified algorithms for a while. We are able to constantly improve and refine our products thanks to research collaborations with the best scientific departments in this field. The result is a great synthesis of science and real-life applications. Get in touch with our consultants and use their know-how to grow your business!

___________________________________________

This three-year research project has been running since 1st January 2018 under the title: ´ TH03010276 – A system for advanced analytics of Big Data based on probability modelling´ and is co-funded by the Technological Agency of the Czech Republic.