Purpose: The purpose of this paper is to develop a quantitative model of problem solving performance of students in the computer-based mathematics learning environment. Design/methodology/approach: Regularized logistic regression was used to create a quantitative model of problem solving performance of students that predicts whether students can solve a mathematics problem correctly based on how well they solved other problems in the past. The usefulness of the model was evaluated by comparing the predicted probability of correct problem solving to the actual problem solving performance on the data set that was not used in the model building process. Findings: The regularized logistic regression model showed a better predictive power than the standard Bayesian Knowledge Tracing model, the most frequently used quantitative model of student learning in the Educational Data Mining research. Originality/value: Providing instructional scaffolding is critical in order to facilitate student learning. However, most computer-based learning environments use heuristics or rely on the discretion of students when they determine whether instructional scaffolding needs be provided. The predictive model of problem solving performance of students can be used as a quantitative guideline that can help make a better decision on when to provide instructional supports and guidance in the computer-based learning environment, which can potentially maximize the learning outcome of students.