Analyzing Competency in Block Credit Prior Learning Assessment: Machine Learning Model Comparisons

• In Event: Machine Learning Methods in Educational Research

Abstract

This study identified factors affecting the average credit hour award among students enrolled in prior learning assessment (PLA) at a public Hispanic-Serving Institution. Multiple machine learning models were compared: linear regression, decision trees, random forests, Support Vector Regression (SVR), k-Nearest Neighbors (k-NN), and Gradient Boosting Machines (GBM) to assess competency in the block credit PLA model that predicts credit hour awards. The results indicated that skill, knowledge, and cognitive ability were key predictors of average credit hour awards. Results further indicated that the SVR model achieved the highest R² value (0.9568) and the lowest MSE (0.8053). It was closely followed by the Linear Regression model with an R² of 0.9559 and MSE of 0.8222.

Authors

• Todd Sherron, Texas State University

• Ruchi Dilip Kukde, Texas State University

• Randall E. Schumacker, University of Alabama

• Lei Xie, Texas State University