Variable Selection in High-Dimensional Logistic Regression Models using a Whitening Approach
Calculate the class-conditional probabilities.
Calculate the weight
Refit the logistic regression with chosen variables
Thresholding on a vector
Thresholding to zero of the smallest values
Thresholding to a given threshold of the smallest values
Variable selection in high-dimensional logistic regression models usin...
tools:::Rd_package_title("WLogit")
Calculate the working response
It proposes a novel variable selection approach in classification problem that takes into account the correlations that may exist between the predictors of the design matrix in a high-dimensional logistic model. Our approach consists in rewriting the initial high-dimensional logistic model to remove the correlation between the predictors and in applying the generalized Lasso criterion.