© Copyright 2019 marcus evans. All rights reserved.
For all enquiries regarding this course contact:
Kamelia Simeonova
E: kamelias@marcusevansuk.com
T: +44 203 002 3172
F: + 44 203 002 3016
Why has there been such a focus in the last few years on the use of machine learning within credit risk?
Machine learning and other kinds of artificial intelligence are permeating more and more areas of life. They achieve better performance more quickly and at lower cost than traditional methods, and they often underlie products that were not previously available. We experience this daily in improved healthcare, cars that warn when we stray out of the lane, Google finishing phrases while we type, etc.
Machine learning has been revolutionizing credit risk predictions as well. Machine learning algorithms can take into account a wide variety of data that are not available to traditional loan officers, and in amounts that would overwhelm a person. Machine learning is also a critical component of fraud detection for online credit applications.
What challenges do established lenders face when competing with fintechs that use machine learning techniques, and how can they solve them?
Fintech companies have been on the forefront of using machine learning for credit risk. One reason is that technologically savvy people are often attracted to the excitement, culture, and potentially large financial returns of working in startup companies. Another reason is that established lenders, which are often part of larger organizations, tend to be conservative about introducing new processes and technologies.
As fintechs capture a larger and larger share of the lending market, though, established lenders are feeling the pressure and starting to incorporate machine-learning technology as well.
An interview with:
Gerhard Mulder
Director and Co-Founder
Climate Risk Services
Machine learning has become an essential tool for evaluating credit risk. This marcus evans course is designed to provide attendees with broad knowledge of how firms can identify where and how to utilise machine learning within their credit modelling and decision-making. Aspects which will be examined include the benefits of machine learning models for lenders, how to improve their explainability, and regulatory issues which might arise from their use, and these will be supported throughout with the use of detailed case studies.
Follow us on Social Media!
About the course
Hershel Safer has more than 35 years of experience developing mathematical models to understand and optimize systems for a variety of applications. He is the founder of Safer Analytics Ltd., a machine-learning consultancy for the finance industry. Over the last decade, he has focused on tackling many of the key applications of machine learning and quantitative modelling within the financial sector. He has worked on machine-learning models to assess credit risk for fintech startups that do invoice factoring and other kinds of business financing. He has also developed AI tools for financial institutions, including process improvement for banks and algo trading for hedge funds. His expertise has been applied to projects executed in the USA, Israel, Europe, and Brazil.
What business issues are created by the use of machine learning within a lender?
When a computer makes credit decisions, people want to understand why it made the decision that it did. On one hand, the objectivity of an algorithm is an advantage, as it obviates concerns about the subjectivity of a loan officer and reasons that a person may make an unwarranted decision.
The flip side is that an algorithm’s decisions may be opaque. Management needs to have confidence that the decisions advance the organization’s goals, fintech investors want to be sure that their funding is used well, regulators want to avoid bias, and loan applicants want to know why they were denied. The need to explain both the overall operation of a decision-making algorithm and the reasons for decisions about specific loan applications affect the choice of algorithm and accompanying tools.
What are some of the regulatory constraints faced by those who wish to use machine learning in credit risk?
Legal and regulatory constraints in many jurisdictions require lenders to be transparent about many lending decisions, especially for consumer lending. Older laws did not foresee the use of machine learning for credit decisions, but they affect our work all the same. These include, in the UK, the Consumer Credit Acts of 1974 and 2006; in the US, the Fair Credit Reporting Act of 1970 and the Credit CARD Act of 2006; and the EU’s Consumer Credit Directive of 2008.
More recently, the GDPR requires changes in how businesses process personal data. In particular, Article 22 requires lenders to be able to explain to loan applicants how their data will be handled, which means that lenders need to understand how their decision-making algorithms work.
How can this course help those in credit risk who are using, or planning to use, machine learning within their departments?
This course is aimed at credit-risk professionals want to understand how to go about introducing machine learning into their organizations. The course will give them the background to:
- Understand how machine learning can benefit the organization;
- Identify the capabilities that they want to hire into the organization;
- Talk with their technical teams about models, algorithms, and automatically generated decisions; and
- Explain to senior management why to allocate resources to machine learning.
As part of his recent work, Hershel has successfully led training workshops on the use of machine learning within credit risk. He has a PhD from MIT in operations research and an undergraduate degree from Yale in applied mathematics and economics.
An interview with:
Hershel Safer
Founder and Consultant
Safer Analytics Ltd.
