Using BIG DATA to Identify Potential Drop-outs
During the FIFA Worldcup Football tournament in 2014 Microsoft has shown the value and accuracy of Big Data Analysis. A team of Microsoft data scientist showed not only proved to be capable of correctly predicting the Final between Germany and Argentina and its 1-0 outcome, but they also correctly predicted the outcome of every single match in the know-out phase. To be able to make these correct predictions the Microsoft team gathered enormous amounts of data. From weather forecasts for the match locations on the day of the game, to the distance between the game location and the hometown of the playing team, to the popularity of the team (measured by social media), the amount of fans presence during the match, their engagement, and of course historic match data. The significance of each data element was updated after every match played, by smart self-learning algorithms.
For universities it is very important to predict the success or failure rate of prospective students. Will they be excellent students, graduating cum laude in just a couple of years, or are they likely to drop out? And while there is still a lot of moral discussion going on about the ethic of refusing potential drop-outs already in the Admission process, almost everybody agrees that it is also in the student’s interest to assist him in making the right choice of program and to address and discuss the potential drop-out risks Therefore, it is important for universities to gather as much information about the potential student as possible, run smart algorithms on them and use the predications that result from this Big Data Analysis in Admission interviews and other communication with the prospective students.
Many universities are still in the experimental phase of setting up Big Data Analysis like this for Admissions, but data items to consider are:
- The timeliness of the application (late applicants have proven to have a higher drop-out rate than early applicants
- The number of programs an applicant shows interest for (the more programs the bigger the doubt, the higher the drop-out risk
- The distance between the hometown of the student and the university (the daily travel might eventually become a drop-out risk)
- The match between the subjects attended in preliminary education and the program of choice
- The high school grade point average
- The historic drop-out rate of students from the same preliminary education institute
- The tuition cost in relation to the family income
- Etc.
For actually using Big Data Analysis in Admissions these are still very early days. Therefore, it is important to learn from each other. And as long as institutions consider the use of Big Data Analysis as a way to optimize the study advise given to prospective students rather than an instrument to beat the competition of other universities, it should not be a problem to actively share knowledge on this topic.
CRM FOR EDUCATION 