My first full-time job was as a Statistical Analyst building credit scoring models at a company that no longer exists. Now virtually all of those models are built by FICO and my old company was absorbed by them years ago.
It was a great first job and one reason was because I certainly learned what it takes to get a loan or a credit card and what NOT to do if you wanted good credit.
I learned that having a credit card or two is good, but having 24 credit cards that are all maxed out is bad.
That’s where some of the art part of building a model comes in.
I talked about the science of building a model in my last blog, but one of the exciting parts of building a model for us analysts is more of an art — finding a new variable that hasn’t been used before.
For example, in building a model to tell if you’re a good credit risk, it may be bad if you have a lot of debt, or it may be a good thing because you’re a high earner and you have a loan for a house and two cars and a boat.
A bigger risk is your debt ratio – you total amount of debt divided by your total amount of credit. If that’s too high and you’re maxed out, that’s bad.
In fundraising, sometimes the variables that are less obvious may tell us that you are more likely to be a major donor or planned giving donor or that you’re more likely to reactivate from being a lapsed donor. The best variables may not be the time since your last gift or the amount of your first gift, but things like the time between your first and second gifts or the percentage increase of the amount your first gift and your last gift.
To me that’s the fun and creative part of building a model. It’s the ART of modeling that makes it beautiful and makes it such an exciting field in fundraising analytics today.