Research on Classroom Interaction Mode Recognition and Evaluation in Higher Education Based on Machine Learning Algorithms

With the continuous expansion of higher education in China, classroom interaction has a direct impact on students' learning outcomes and the cultivation of key abilities. Traditional methods of evaluating classroom interaction are often subjective and have limited assessment effectiveness. Machine learning algorithms, on the other hand, can provide more objective and accurate evaluation results by analyzing large amounts of data. This paper constructs a classroom interaction mode evaluation model based on SVM (Support Vector Machine), using the Gaussian RBF kernel function for data processing, and proposes a new evaluation system. In the experimental section, 384 data samples from classroom videos, surveys, and expert ratings were collected. After data cleaning and preprocessing, the SVM model was used for training and prediction. By comparing the classification accuracy of three algorithms—BP Neural Network, Decision Tree, and SVM—on multiple datasets, the results show that SVM outperforms the other two algorithms on all datasets, with an accuracy rate of 97.06% on the Data 2 dataset. Additionally, the experiment also compared the time consumption of different algorithms during training and testing, revealing that SVM has the shortest testing time and high time efficiency. Overall, the SVM algorithm not only performs excellently in terms of accuracy but also demonstrates significant advantages in time efficiency, providing an efficient and reliable tool for evaluating classroom interaction modes in higher education.