Analyzing log files to predict students' problem solving performance in a computer-based learning environment

Abstract

This study investigates whether information saved in the log files of a computer-based tutor can be used to predict the problem solving performance of students. The log files of a computer-based physics tutoring environment called Andes Physics Tutor was analyzed to build a logistic regression model that predicted success and failure of students’ problem solving from their past interactions with the computer-based tutor. The logistic regression model developed in this study was able to correctly identify about 70% of the observed problem solving performance. The 10-fold cross-validation and the Receiver Operating Characteristic (ROC) curve analyses suggest that the developed logistic regression model can predict students’ problem solving performance on unseen new problems with a similar accuracy in the future.