Colleges Mine Data to Predict Dropouts.

The article discusses the use of data to improve college student retention. Such a system uses a computer algorithm and has been implemented at Purdue University to alert students if they are performing well in their class. The algorithm can allegedly predict which students will drop out and alert school staff of the situation. A similar system at the State University of New York at Buffalo looks at seven variables to assess students. The flaws in the prediction models include overlooking people who are at-risk and sometimes students can remedy their own situations.