Researchon Application Of Outliers Processing In The System Of Teacher’s Performance Evaluation

Teachers’ performance appraisal has important influences on reforming the educational personnel system, developing the teaching staff, and progressing the education cause. Based on the particularity of teaching jobs, a teacher’s performance is often assessed by other teachers and the students. In coworkers’ mutual rating, they probably degrade the marks to each other because of the competition, or upgrade the marks to each other because of the interpersonal relationship. In students’ evaluation, they probably mark perfunctorily, or give a hostile poor commentary to the severe teachers. Therefore, it is necessary to deal with the extremely data of teacher’s performance evaluation in order to get the fair results.This paper mainly includs the following aspects:Firstly, the reasons of Outliers in teachers’ performance appraisal system will be analyzed, and the Outliers will be classified according to their different characteristics.Secondly, it analyzes the common statistical test methods of the Outliers. And also, it analyzes the shortcomings of these methods based on the characteristics of teachers’ performance appraisal.Thirdly, it discusses how to diagnose and correct the outliers based on different classification.It puts forward to the way of using quartile robust statistics to diagnose the outliers of teachers’ mutual assessment. It also suggests how to correct the outliers by introducing the parameters of deviation and dispersion in order to avoid the influences on the small sample set of teachers’ mutual assessment by directly deleting the outliers.The outliers of students’ assessment can be divided into perfunctory data and prejudice data. Based on the different traits, it analyzes the perfunctory evaluation data by using the correlation coefficient between evaluation indicators, and tests the prejudice data by using Pauta criterion. According to the characters of big sample set, theOutliers will be directly removed.Finally, it deals with a set of data including 15 teachers for a given semester from a secondary vocational school to illustrate the effectiveness of the algorithm applied in the teachers’ evaluation system. In the case, it deals with the appraisal data from the coworkers and nearly 600 students. The results indicate that there is no significant difference between the coworkers’ and the students’ appraisal. In brief, this paper comes up with certain methods to appraisal the teachers’ performance objectively and impartially.