| In recent years,the art of education has blossomed and Art Grading has entered the fast lane.But,the process and form of Art Grading are always at the traditional level.With the deep penetration of the Internet and the continuous development of social art test,"Internet + Art Grading" has gradually moved toward people’s perspective and has been paid attention and got promotion of the industry.However,domestic research of Network Art Grading lacks existing specification and theory.Data research of Network Art Grading is still a problem that is worth discussing to provide valuable experience on the development of China’s Network Art Grading and the basis for government departments to formulate relevant policies.Models and algorithms are used to analyze the data of Network Art Grading in this article.Specific works are as follows:(1)The theories of SVM and LR in machine learning classification algorithms are introduced.Combined with examples,the two algorithms are used to classify the candidates with excellent and non-excellent.(2)The theories of SVR and CART in regression algorithms are introduced.The model of test data is established by using the different kernel functions of SVR,and the conclusion that the Gaussian kernel function has a better prediction effect on examination results is reached.At the same time,some characteristics of the test data are selected,and the multiple linear regression and tree regression are used to analyze the modeling.It is concluded that the tree regression has a better prediction effect.(3)Combined with the instance,the missing value of tests data is estimated by using matrix decomposition algorithm,which provides a reliable method for the evaluation of excellent training institutions when there is a few missing data.L2 regularization item that can help get a better generalization ability is applied to avoid over fitting phenomenon.(4)The area of Network Art Grading is grabbed by web crawler of R language and a time series model of SPSS is established to produce analysis and visualizations of test data.At the same time,the distribution of specialty,grade and age of the Network Art Grading of 2017 in Zhejiang province is given,which provides reference for the government and development of the industry.(5)The work of this paper is summarized and the future research direction is put forward. |