frobstepwise {gausscov} | R Documentation |
robust stepwise selection of covariates
frobstepwise(y, x, cn, alpha, nu = 1, kmax = 0, intercept = TRUE, chkintercept = FALSE, kexk = 0, sig = 0)
y |
Dependent variable |
x |
Covariates |
cn |
Parameter fir Huber's psi-function |
alpha |
The P-value cut-off |
nu |
The order statistic of Gaussian covariates used for comparison |
kmax |
The maximum number of included covariates |
intercept |
Logical to include intercept |
chkintercept |
Logical to include or exclude intercept dependent on the P-value |
kexk |
The excluded covariates |
sig |
Scale value of dependent variable |
pv pv[[1]] the included covariates, the P-values; pv[[2]] coefficients of robust linear regression; pv[[3]] residuals; pv[[4]] scale.
data(boston) bostonrob<-frobstepwise(boston[,14],boston[,1:13],1,0.01,15,intercept=TRUE) bostonrob[[1]] bostonrob[[2]] plot(bostonrob[[3]])