Risk Prediction
The assessment of the solvency of bank client companies is influenced by many events occurring in the industry, and most of them are reflected in open news sources. Assistance to bank analysts in analyzing the heterogeneous risks of a borrowing company according to the news flow consists in automating the process of collecting and analyzing news by creating a solution to tagging and categorizing news based on a given set of risks.
Input Data
- News Seed on the selected company-borrower
- A set of risks specified by the bank
- Markup news according to risks from analysts
Output Data
- The model estimates the probability of existence for a given risk of the borrower to the preceding News
Main Results
Integration into the business process of the loan portfolio analysis department of the bank:
- Automation 60% of the staff searching for relevant information
- Reduced analyst load up to 70% in different risk categories
- Increase the accuracy of department forecasts by 15%
- Increasing the number of detailed reports on the portfolio in 2 times
- Compiling semantic core for 80% of client companies
The final quality of the semantic core selection:
Quality criterion | Quality |
---|---|
Name and its variants | 75 % |
Names of the leaders | 70 % |
Affiliations | 68% |
Industry Terms | 80 % |
Quality of the constructed risk analysis system
Quality criterion | Quality |
---|---|
F1-score | 97 % |