Search

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.

The article author discusses the power of big-data and prediction algorithms in making accurate predictions of which tweets will be retweeted often, with consideration on their boundaries and limitations relating to correlation, causality problem, and the paradox of predicting what is interesting.