| MDCSIS {MFSIS} | R Documentation |
Martingale Difference Correlation and Its Use in High-Dimensional Variable Screening
Description
A new metric, the so-called martingale difference correlation, to measure the departure of conditional mean independence between a scalar response variable V and a vector predictor variable U. Our metric is a natural extension of distance correlation proposed by Szekely, Rizzo, and Bahirov(2007), which is used to measure the dependence between ´ V and U. The martingale difference correlation and its empirical counterpart inherit a number of desirable features of distance correlation and sample distance correlation, such as algebraic simplicity and elegant theoretical properties.
Usage
MDCSIS(X, Y, nsis = (dim(X)[1])/log(dim(X)[1]))
Arguments
X |
The design matrix of dimensions n * p. Each row is an observation vector. |
Y |
The response vector of dimension n * 1. |
nsis |
Number of predictors recruited by MDCSIS. The default is n/log(n). |
Value
the labels of first nsis largest active set of all predictors
Author(s)
Xuewei Cheng xwcheng@csu.edu.cn
References
Szekely, G. J., M. L. Rizzo, and N. K. Bakirov (2007). Measuring and testing dependence by correlation of distances. The annals of statistics 35(6), 2769–2794.
Shao, X. and J. Zhang (2014). Martingale difference correlation and its use in high-dimensional variable screening. Journal of the American Statistical Association 109(507),1302–1318.
Examples
n=100;
p=200;
rho=0.5;
data=GendataLM(n,p,rho,error="gaussian")
data=cbind(data[[1]],data[[2]])
colnames(data)[1:ncol(data)]=c(paste0("X",1:(ncol(data)-1)),"Y")
data=as.matrix(data)
X=data[,1:(ncol(data)-1)];
Y=data[,ncol(data)];
A=MDCSIS(X,Y,n/log(n));A