Machine (Data Mining) Credit Scoring
Background
One of the earliest uses of data mining methods was to set up computer systems to make credit decisions for banks and similar financial institutions. In this case, assume you have been hired by a small community non-profit credit union to investigate the feasibility of replacing their current credit scoring system (done by human loan officers) with a computer data mining algorithm. Note that a non-profit community credit union has as its primary mission to support the financial/economic needs of its community and may be less “profit driven” than a regular bank. On the other hand, if the credit union loses money, it won’t be able to stay open for long and then it won’t serve anybody so it still has to be responsible about the loans it chooses to make.
Due to the lateness, and the oddness, of this semester we’re going to do this case differently. So 5 points of the 10 you usually get will be to do the technical analysis — as a homework problem (in Excel, with answers in yellow boxes). The second 5 points will be an essay, discussing your take on the pros and cons of using machine learning algorithms to support and/or make business decisions.
The file “Machine Credit Scoring” contains the data for this case and is set up as a homework problem. The data set has customer information on 425 previous customers of this credit union and shows if the loan committee rated each as a “high” or a “low” credit risk. (Here “low” is good; the credit union will be more likely to make loans, and to make larger loans, to customers who are at “low” risk of having problems repaying these loans.) In the second worksheet, are 10 new loan applicants. The technical goal of this case is to develop the best classification model that you can, and then to use that model to classify each of the new applicants as being a “high” or “low” credit risk. Follow the directions, and put your answers in the spreadsheet. Submit that as one of your submissions for the Chapter 10 case.
As for the other 5 points, discuss, from your own point of view some of the pros and cons of these machine learning algorithms. Are they a competitive tool that every company, large and small, must learn and apply if they wish to remain in business in the 11st century? Will they lead to the collapse of society? ( assuming you’ll fall somewhere in-between.) Below are a couple of short “greatest thing ever” type articles, and a couple of “Danger! Danger!” articles for you.
Sections should include:
- Introduction (set the stage, and at least tentatively take a position),
- Pros of Using Machine Learning Algorithms,
- Cons of Using Machine Algorithms, and
- Conclusion (where you should restate your position)
Final work to be 3-5 (double spaced) pages.
From: https://www.consumer.ftc.gov/articles/0347-your-equal-credit-opportunity-rights
The Federal Trade Commission (FTC), the nation’s consumer protection agency, enforces the Equal Credit Opportunity Act (ECOA), which prohibits credit discrimination on the basis of race, color, religion, national origin, sex, marital status, age, or because you get public assistance. Creditors may ask you for most of this information in certain situations, but they may not use it when deciding whether to give you credit or when setting the terms of your credit. Not everyone who applies for credit gets it or gets the same terms: Factors like income, expenses, debts, and credit history are among the considerations lenders use to determine your creditworthiness.
Resources Regarding Machine Learning Pros and Cons
Pros/Advantages of Machine Learning Algorithms
An “oldie but goodie”. This one is referenced in the textbook: http://akira.ruc.dk/~bulskov/undervisning/E2003/data_mining.pdf
https://www.wordstream.com/blog/ws/2017/07/28/machine-learning-applications
Cons/Limits of Machine Learning Algorithms
https://www.brookings.edu/research/fairness-in-algorithmic-decision-making/
https://www.amazon.com/Weapons-Math-Destruction-Increases-Inequality/dp/0553418815
