Research On The Quality Of Classroom Teaching Evaluation In Universities Based On Support Vector Machines

At present, classroom teaching is the main teaching form in universities. It is the foundation of university teaching and plays a very important role in the teaching process. The establishment and implementation of classroom teaching quality evaluation system not only help develop teaching theory, also ensure the evaluation of the quality of classroom teaching smoothly and play an effective role of the classroom teaching activities.The traditional teaching quality evaluation which is only the students’ participation has some advantages, but there are still some problems. For example, students have a prejudice against teachers, or focus only on the evaluation results but not the teaching process and so on, so that the evaluation results appear error. After Support Vector Machines(SVMs) are introduced into the teaching quality evaluation, students, peers and leaders are involved in the evaluation, and it can not only avoid errors caused by students’ factors but also fully reflect the teaching process of teachers. In addition, teaching quality evaluation is a multi-class classification problem, so SVM multi-class classification algorithm is used to forecast the results of classroom teaching quality evaluation in this paper. To sum up, the main work is as follows:(1)The significance of the traditional evaluation methods and the disadvantages of traditional evaluation methods are analyzed and summarized. According to the specific needs and principles of constructing evaluation system, the evaluation system of classroom teaching quality is established in this paper. In view of non-linearity between evaluation indexes, SVMs are applied to teaching quality evaluation to solve possible problems in teaching quality evaluation.(2)Several SVM multi-class classification algorithms are introduced, including the binary tree SVM multi-class classification algorithm. A new improved algorithm is proposed since the binary tree SVM multi-class classification algorithm exists the problem that only create partial binary tree. The improved algorithm uses the generation strategy of complete binary tree and definitions of clustering class distances, so that the binary tree achieves complete or nearly complete state. Finally, the simulation experiment is carried out on UCI data sets to prove the effectiveness of the improved algorithm.(3)The classroom teaching quality evaluation model for university is established by using the improved algorithm which based on binary tree SVM in this paper. To evaluate the teaching quality of a university in Shandong Province, we fill out the evaluation questionnaire, statistics and collect multiple sets of data. The data sets are tested, and the experimental results are analyzed in the MATLAB environment. Compared to the SVM algorithm and the binary tree SVM algorithm, the improved algorithm has obvious advantages in prediction accuracy and efficiency.