Research On Outlier Detection And Its Application In Teaching Evaluation Model

Teaching evaluation is a process of measuring,analyzing and evaluating the teaching process and results based on the teaching objectives,according to scientific evaluation criteria and using effective technical means.Teaching evaluation plays an important role in the process of school teaching.The results of students'evaluation of teachers play an important role in the evaluation of teachers'professional titles,the construction of teachers'ranks and the promotion of teaching reform.However,due to various reasons,there are some students who perfunctorily,deliberately praise or maliciously evaluate teachers who teach severely and earnestly.In this way,some data outliers will appear in the teaching evaluation data.If these outliers are not properly handled,they will eventually have a negative impact on the objectivity and impartiality of teachers'assessment results.This paper takes"outlier detection and its application in teaching evaluation model"as the research topic.Through outlier detection,processing and data optimization in teaching evaluation data,the adverse impact on the assessment results is weakened,and the evaluation results are more fair and fairThe following research will be completed according to my work practice:Firstly,In view of the characteristics of complex learning situation and different density distribution of students'teaching evaluation data,a local Outlier Factor(LOF)detection algorithm based on relative density is proposed in the teaching evaluation data,and the LOF algorithm is designed and implemented.Secondly,Selecting a full-time teacher's teaching evaluation data of different classes and curricula from the database of the teaching evaluation system as a sample set for LOF outlier detection,Based on the analysis of the test results,the outlier correction formula is formulated to correct the outlier.The results of variance analysis show that the variance and standard deviation of the corrected data set are smaller than those of the corrected data set,and the stability of the corrected data set is better than that of the corrected data set.Thirdly,according to the actual work,this paper systematically analyses and designs the data processing requirements of the teaching evaluation system.Then 960students are randomly selected from the database to detect and correct the outliers of5760 teaching evaluation data of 16 teachers.By comparing the trend charts of the scores of the students before and after the correction with those of the supervisors,the linear trend line formula and the correlation coefficient R~2 values are analyzed.It fully illustrates that the fitness between the corrected students evaluate teacher's score and the supervisors evaluate teacher's score is better than that before the correction.It is concluded that the LOF outlier detection algorithm and its correction have better application effect in the teaching evaluation system.