Evaluation of Student Performance Based on Learning Behavior with Random Forest Model

With the rapid development of online education, researching on the Education Data Ming (EDM) has been studied extensively in recent years. The study of online learning behavior data has become an important aspect of EDM. In contributing a deeper verifying of applying with EDM, this study firstly preprocessed online learning behavior data and analyzed the correlation between online learning behavior data and student performance. By inputting the selected features into LightGBM (Light Gradient Boosting Machine) model, random forest model, support vector machine regression model and linear regression model, we had mined the correlation coefficient between the behavioral features. By comparing the evaluation indicator of data mining, findings showed the algorithm priority and the random forest model was the best data model. Specifically, applying the advanced random forest model for verifying, the behavior data of about 300 students undertaking a major in tourism management was meticulously examined and validated. The results showed that the prediction accuracy of random forest model was up to 98.15%, which further confirmed the performance of random forest algorithm. This research combined student behavior data and student scores with data mining technology, and through sufficient experimental verification, the experimental process, methods and results were validated. The study suggests that random forest algorithm can be applied to analyze learning behavior data generated by other learning platforms to predict students' final performance and provide judgment basis for reducing students' dropout risk based on the results.