Сustomer problem analysis: defaults and loan defaults?


Application/antifraud scoring


High rate of loan defaults. The need to assess the creditworthiness of a potential borrower as part of the loan application process.


AI credit scoring. It is an intelligent module for assessing the creditworthiness (credit risks) of a potential client based on deep machine learning and neural networks.
It can predict the probability of performing credit payments over the next two years with an accuracy of 90%.


The AI module helps analyse information about the bank’s customers using artificial intelligence and machine learning.

It can examine data from social media and databases, analyse potential borrowers’ online activity and even take their behavioural patterns into account.

The AI module reduces the human factor in decision making.

Analysis of non-standard borrower information, i.e. more accurate prediction of client default, compared to classical scoring systems.

Technology stack

  • Python
  • TensorFlow
  • Keras
  • Sklearn
  • XGBoost