The Application Of Multivariate Statistical Analysis In College Classroom Teaching Assessment

With the rapid development of China's Higher Education ,the quality of teaching institutions of higher learning is receving more and more people's attention. Higher teaching quality assessment is the basis for dynamic management, but also to achieve total quality management education, an important guarantee. The contents of the thesis are as follows:(1) The literatures on classroom teaching assessment in recent years are summarized and the methods used in factor collection and the transition of analysis methods are reviewed. Some comments and suggestions are provided. Hope for the college teachers in classroom assessment study to provide reference. Meanwhile, the classroom assessment questionnaire in this paper is analyzed simply.(2) In view of Mahalanobis distance not taking into account the relative importance of each indicator variable, the weighted Mahalanobis distance discriminant rule is used in the college classroom teaching assessment.Firstly,we roughly cluster all the samples through indicators of the overall score,then calculate the mean and covariance matrix of each type and with each category of mean and covariance matrix seek weighted Mahalanobis distance value. According to the size of the value of the weighted Mahalanobis distance,we re-classify all the samples. The conclusion shows that the method based on the weighted M-distance discriminant rule could separate the higher cases better.(3) In view of the principal component analysis are linear comprehensive evaluation, with certain limitations, a non-linear multi-tiered evaluation model is given. Firstly, all questionnaire data of classroom teaching assessment is carried on the Box-Cox nonlinear transformation. Then, the transformed data is carried on the target cluster analysis and the multi-factor system is established. Principal component analysis is carried on in each index. The multivariate hierarchical model for the classroom teaching assessment based on Box-Cox transformation is established. The conclusion shows that the significance of index classification_which is given by principal component analysis using the Box-Cox transformation is more obvious, and at the same time the higher cases could be separated better.