The Prediction Modeling System for Monitoring Elementary Students’ Mathematics Progress in Online Curriculum-Based Measurement

The curriculum-based measurement (CBM) is a data-based individualization method that monitors students’ performance and improvement over time. The purposes of this research were to develop prediction models—including ordinary least squares linear regression (OLS), Gaussian Naive Bayes, Bayesian Networks, and Random Forest—within a web-based CBM system, and to investigate their effectiveness in predicting elementary students’ mathematics performance. A total of 92 fourth-grade students participated in the study. They used mobile devices to complete CBM probes over an eight-week period. Performance metrics were analyzed to evaluate the error rate between predicted and observed scores. Overall, the results showed that OLS and Bayesian-based models were effective in predicting elementary students’ mathematics performance. Moreover, the findings indicated that distinct growth patterns still existed across different classes.