For the seventh installment of The Weight List, Capture Higher Ed’s data science podcast, Dr. Thom Golden and Dr. Brad Weiner take on an excellent question submitted by a listener about predictive modeling.
The question: How can a university use historical data to build a predictive model, and then apply that forward to a market that might be completely new or in a developmental stage?
According to Thom and Brad, this is when a statistical algorithm must be a really good archeologist. Because a university’s historical data includes all kinds of “artifacts” that help an algorithm understand what’s been happening at that institution.
“Those patterns might not be detectable to the human eye, or they may not be detectable through normal statistical methods, but they are detectable,” Brad says. “And no matter where you’ve been in the past, the assumption is that your future is going to look something like your past.”
Listen to Thom and Brad’s full discussion on how a “rear facing” prediction applies to new student markets on The Weight List, Episode 7.