Statistical Analysis Of Teaching Evaluation Data Of Contemporary College Students

It is a common method of teaching evaluation that students grade teacher,and it is widely used in various colleges and universities.It is one of the important methods to check the quality of teaching.The analysis of teaching evaluation data can help to discover the problems exposed in teaching,so that we can take measures to guide teaching and practice in the right direction.The article uses the data of examination grades of 2011-2014 and teaching evaluation data of 2011-2017 of Chongqing University to analyze the relationship between the scores of students' evaluation of teaching and students' grades.Firstly,ANOVA was used to compare the differences of students' evaluation of teaching under different type of course.It was found that students' evaluation of teaching differed greatly in the three major sections of professional courses,public courses,and general courses.Therefore,in the subsequent analysis,the data was divided into three major sectors.After the correlation analysis,we found the following conclusions.In total,137 courses were found to be correlated to different degrees,among which 79 were specialized courses,accounting for 2.8% of the corresponding courses.The course with the highest correlation coefficient was finance,with the correlation coefficient r=0.7462.There are 23 public courses,accounting for 16.7% of the corresponding courses.The courses with the highest correlation coefficient are college English(4)of the school of automation,r=0.661.There were 11 general courses,accounting for 5.8% of the corresponding courses.The course with the highest correlation coefficient was introduction to popular culture(r=0.613).Analysis the cause of the correlation,on the one hand,active and positive reason for teacher teaching earnestly,good effect,causes the student support,and excellent teaching standards,student performance in correlation analysis showed positive correlation,on the other hand,the reason may be the reason for the negative,negative,is a score "conspiracy",teachers and students at the expense of the damage to the quality of teaching,make teachers and students get score on the "win-win",these very good or very bad reasons can cause course grade relevance to the results.In the paper,the variance of each course score is compared with the variance of the population of the corresponding class,which is the main basis to infer the correlation between the students' teaching scores and grades.The second part carries out regression fitting for the data with significant correlation.We use python to make an interactive interface.On the basis of establishing the linear model of least squares estimation,the index and visualization of the target data model are realized,and the results of correlation analysis are visually displayed to the greatest extent.Then,in view of the robustness risk of OLS(least squares estimation)linear model,theil-sen estimation was introduced to establish a linear model,and examples were selected from the data for comparison.The advantages of theil-sen model were discussed from the aspects of robustness and fitting accuracy,and the supplement of OLS model in non-normal data modeling was realized.Finally,the logarithmic function model and polynomial regression are used to carry out nonlinear fitting respectively.The fitting results show that linearization of the nonlinear regression problem in the form of logarithmic function model can further improve the goodness of fitting to a certain extent.The three methods of least squares estimation,Theil-Sen estimation,and nonlinear regression are used to fit the data,which can better reflect the development trend and related conditions of each branch of data.It is benefit to improve the quality of teaching.