Research On Anomaly Detection And Visualization Technology In College Student Physical Fitness Test Management System

Physical fitness test(PFT)of students is an important task of physical education in college.However,it is still difficult in using traditional method like Excel to process large amounts of PFT data.The college student physical fitness test management system is a B/S system designed for the PFT workflow.It can greatly improve efficiency in PFT work,avoid the errors caused by semi-manual operation.It can also analyze PFT data to discover problems and analyze the development trend in PFT by using machine learning and data mining techniques.In addition to the basic functions of the management information system,it also has multiple functions such as importing score,calculating score,reporting results,and notice¬ification.These functions can solve the most disturbing problem for teachers in physical fitness testing.At the same time,the system also has two core functions:abnormal data detection and data visualization.Abnormal data that often occurs for various reasons in PFT can degrade the credibility of the overall data.Abnormal data may reduce the fairness of the test and impede data research.According to the characteristics of PFT data and combining with the anomaly detection method based on distance,density and clustering,an anomaly detection fusion algorithm for PFT(ADF-PFT)is proposed.ADF-PFT integrates the k-means clustering result,the Euclidean distance between the current data and the historical data,and the density through specific weights.The fusion result is a measure of the degree of abnormality.Through experimental verification,it is concluded that the algorithm can find out abnormal data more accurately,efficiently and with less errors in large amounts of PFT data,which can assist the college sports teachers to deal with abnormal data.With the aid of web-based visualization tools,applying data visualization techniques to the PFT management system can play an important role in the analysis of PFT data.The pretreatment scheme for physical test data is studied according to the basic processing flow of data visualization.Especially for missing value in data,a comprehensive processing strategy is proposed:when a few values are missing,using stochastic filling method based on Naive Bayes classification;when the more missing values in data,using weighted mean filling method based on KNN;when a large number of missing values in data,delete data directly.In addition,the system also provides multiple visualizations of application scenarios.We found and proved the difference in the overall scores of male and female students' PFT score through the scientific analysis of visual images.The total score of female students is always higher than that of male students.Ultimately,visualized image can indicate that the difference is caused by a large difference in the passing rates of force evaluation indicators.