Machine Learning Regression Analysis of EDX 2012-13 Data for Identifying the Auditors Use Case

Mark Mueller, Greg Weber
2017 International Journal on Integrating Technology in Education  
Predictive models are able to predict edX student grades with an accuracy error of 0.1 (10%, about one letter grade standard deviation), based on participation data. Student background variables are not useful for predicting grades. By using a combination of segmentation, random forest regression, linear transformation and application beyond the segmented data, it is possible to determine the population of the Auditors student use case, a population larger than those students completing courses
more » ... completing courses with grades.
doi:10.5121/ijite.2017.6301 fatcat:clcc3wrjyfbqzpshnpnynvdiii