Webinar:
"Development Approaches ML Models for Credit Risk Analysis Tasks"
"Development Approaches ML Models for Credit Risk Analysis Tasks"
Date and time
TBA
11:00 - 12:30 AM
11:00 - 12:30 AM
Online
In this webinar, we'll delve into fresh approaches to developing ML models specifically tailored for analyzing credit risks. We'll kick things off with an overview of typical tasks and the tried-and-true algorithms used to tackle them. From there, we'll explore potential growth areas and present various avenues for advancement, including the application of neural networks. To wrap up, we'll take a close look at implementing a common model training and validation framework within the Kolmogorov AI platform.
The information will be useful for both Data department managers and Data Science specialists.
The information will be useful for both Data department managers and Data Science specialists.
AGENDA
1.Credit risk modeling process.
2.Approaches in credit risk modeling:
- Classical methods: logistic regression and gradient boosting.
- Modular principle: PD modeling in credit cards, case study of "Bank 'Otkritie.'"
- Advanced methods: recurrent neural networks.
- Utilization of features from graphs.
3.Challenges encountered during ML model training.
4.Demo. Implementation example on the Kolmogorov platform.
KEY SPEAKERS
- Polina OkunevaThe webinar's author. Head of Analytics in Finance and Risk.
Holds extensive experience in projects utilizing the AB-testing methodology across various industries. - Timofey PribylevSeasoned Analyst specializing in Machine Learning.
REGISTRATION
Kindly complete the form provided below.
The link for translation will be sent to
your email the day before.
The link for translation will be sent to
your email the day before.
We do not host your personal information.
By clicking the "Register" button, you consent to the processing of the personal data you have entered in the form and its transfer to the organization's representative.
The organizers reserve the right to refuse participation in the event.