Research And Application Of Linear And Nonlinear Correlation Coefficient Between Classes

In the era of large data,the data have the complex characteristics of mass,diversity and so on.The correlation analysis of data has begun to attract people's attention.Nothing in the world can exist independently.They are more or less related,and the correlation coefficient can reflect the correlation degree between things.Therefore,it is very meaningful to study correlation coefficient.This article is to further explore the related analysis in the background of large data,and get the following results:?1?The study of correlation coefficient is usually based on ordinary deterministic set,such as Pearson correlation coefficient andpartial correlation coefficient.Inspired by the scholars on the correlation coefficients of fuzzy sets,combining the Pearson correlation coefficient and rough set,a method of calculating the correlation coefficient based on the rough sets,the applicable conditions of Pearson correlation coefficient,the method of characterization is the linear relationship between the equivalence classes of rough set.This paper not only proves the correctness of the method,but also proves the validity of the method by an example.?2?The traditional statistical correlation analysis is used to characterize the linear relationship among variables.Mutual information based correlation analysis is used to describe the nonlinear relationship between two variables.Distance based correlation analysis is used to describe the nonlinear correlation of high-dimensional data.The linear and non-linear relationship between variables are studied,and this paper is a study of the correlation coefficient between classes,found Hilbert-Schmidt Independence Criterion?HSIC?is a nonlinear relationship between the study variables,and is applicable to a wider data set type,no longer limited in rough set.Based on HSIC's empirical estimation?HSIC₀?,a method of measuring the nonlinear correlation between classes and classes based on class labels is proposed.In this paper,three types and six sets of real data sets are selected,and four kernel functions including linear kernel,polynomial kernel,RBF kernel and Sigmoid core are used to verify it.The results show that the method is feasible.In conclusion,this study is the correlation coefficient between classes,the linear correlation coefficient is set to solve the uncertainty of rough set,nonlinearcorrelation coefficient is applied to any set of effective and real data sets show that the proposed method has very good practical significance.