We are exploring areas of learning analytics using a data set from San Diego State University students in a particular statistics course. Often the university targets specific student groups such as first generation or commuter students and encourages/requires them to take supplemental statistics courses to remediate and provide them the support needed for the rigor of college courses with the ultimate goal to achieve higher grades. Using the students in optional, supplemental courses as the group of interest, we will analyze the effectiveness of these supplemental courses on student performance. Often in educational data analysis, there is little to no control over how the subjects are selected. To reduce the bias between those students in the supplementary courses and those not, we apply a novel random-forest based method for propensity-score matching of students in this observational study setting.