| IPODFUN {leapp} | R Documentation |
compute the iterative penalized outlier detection given the noise standard deviation sigma
Description
Y = X beta + gamma + sigma epsilon estimate k by 1 coefficients vector beta and N by 1 outlier indicator vector gamma from (Y,X).
Usage
IPODFUN(X, Y, H, sigma, betaInit, method = "hard", TOL = 1e-04)
Arguments
X |
an N by k design matrix |
Y |
an N by 1 response vector |
H |
an N by N projection matrix X(X'X)^-1X' |
sigma |
a numeric, noise standard deviation |
betaInit |
a k by 1 initial value for coeffient beta |
method |
a string, if "hard", conduct hard thresholding, if "soft", conduct soft thresholding, default to "hard" |
TOL |
a numeric, tolerance of convergence, default to 1e-04 |
Details
The initial estimator for the coefficient beta can be chosen to be the estimator from a robust linear regression
Value
gamma |
an N by 1 vector of estimated outlier indicator |
ress |
an N by 1 vector of residual Y - X beta - gamma |
Author(s)
Yunting Sun yunting.sun@gmail.com, Nancy R.Zhang nzhang@stanford.edu, Art B.Owen owen@stanford.edu
References
She, Y. and Owen, A.B. "Outlier detection using nonconvex penalized regression" 2010