Student Learning Behavior Analysis and Feedback Algorithm based on Intelligent Model

The learning behavior of students in online learning environments is difficult to quantify and evaluate, and there is a lack of effective personalized feedback mechanisms. This paper models students' learning behavior, reveals the inherent laws and patterns of students' learning behavior, and predicts students' mastery of each knowledge point and learning performance. Based on the student learning behavior model, personalized learning resources and recommended content are provided to students. The CNN-LSTM model has an average accuracy of 97.79% in predicting students' learning behavior and an average accuracy of about 95.77% in predicting knowledge mastery. In addition, the average score (95.9) of the personalized learning group is higher than that of the unified learning group (86.3), and the average completion rate (98%) is also higher than that of the unified learning group (91.04%). This study more accurately predicts students' learning outcomes and mastery of knowledge points.