The Research On Functional Principal Component Analysis For The Blood Test Data Of Coronary Heart Disease

Functional data analysis(FDA)transfer the discrete data into whole data by using a function,and then we can use a series of mathematical methods which usually address the continuous data to deal with the original sample,so as to achieve its potential characteristics and properties.Functional data analysis method makes the continuous and the discrete data closely linked,the continuous process for the discrete data be realized,and obtains the potential high dimensional function instead of the original low dimensional data,which also can obtain more abundant and reliable information,and it can reflect the changes of the data which can not be directly expressed.Traditional principal component analysis(PCA)can process those data that have a certain intrinsic correlation between the original variables into a few and relatively independent of several new variables.If it is the functional data,the measured data may be in very dense time point,at this time,the different observations of individuals may have different points of time,or there are some missing values.It will induce the inaccuracy of the results when use the PCA method to deal with the problem.Therefore,you can use the functional principal component analysis(FPCA)to solve this problem when the data is the functional data.This paper presents a method of FPCA for blood test data of coronary heart disease.First,we transfer the blood test data of coronary heart disease into functional data by means of the idea of FDA.Then,we use FPCA to reduce the dimensions of the original test index,and compare its results with the methods of PCA.The experimental results show that,compared with PCA,the indexes reduction has better effect by using the method of FPCA and the new indexes can contain most information of original index,which is of great significance to the diagnosis of coronary heart disease.