By Rob Mitchum // January 16, 2014
[The Eric & Wendy Schmidt Data Science for Social Good Summer Fellowship is now officially accepting applications for 2014 at at dssg.uchicago.edu. The deadline for fellows and mentors is February 1st. During the application period, we will cross-post excerpts from articles about last year’s DSSG projects.]
For high school graduates, choosing the right college depends on many different factors. Academics, cost, distance from home, athletics, and social factors all play a role in choosing which schools to send applications. But students and their families often have incomplete information about the colleges they are qualified academically to attend or those that will provide the necessary financial aid. When a student settles for a college beneath their qualifications, the phenomenon is called “under-matching.”
The 2013 Data Science For Social Good team of Min Xu, Edward Su, Nihar Shah, and mentor Michelangelo D’Agostino used data analytics to tackle this educational problem. Working with the public school district of Mesa, Arizona, the team developed a model to predict those students at high risk of undermatching.
Once we’ve learned which factors were predictive of kids who undermatched in the past, we’re ready to predict the futures of current students – kids who have yet to graduate that Mesa could target for intervention. This is the whole point of predictive analytics – being able to make good guesses about someone’s outcomes before they happen, when you can still change them.
We take those factors that were predictive in the past and use them to construct a model, so that we can take information about a graduating senior who has yet to apply to college, plug it into the model, and estimate how likely she is to undermatch.
You can read the rest of Edward Su’s blog about the project at the DSSG website, or view more information about the methods used at the project’s GitHub repository.