This function dose k-fold cross-validation for the Network-based logistic regression and returns a pair of lambda1 and lambda2.
CV.NetLogistic(X, Y, lamb.1 = NULL, lamb.2 = NULL, r = 5, alpha.i = 1,
folds = 5)a matrix of predictors.
a vector of the binary response.
a user-supplied sequence of lambda1. Tuning parameter lambda1 imposes sparsity. If it is left as NULL, a default sequence will be used.
a user-supplied sequence of lambda2. Tuning parameter lambda2 controls the smoothness among coefficient profiles. If it is left as NULL, a default sequence, c(0.1, 1, 10), will be used.
the regularization parameter in MCP, default is 5.
by default, the program use the lasso for choosing initial values of the coefficient vector. alpha.i is the elastic-net mixing parameter, with \(0 \le alpha.i \le 1\). alpha.i=1 is the lasso penalty, and alpha.i=0 the ridge penalty. If assign alpha.i to be -1, program will use zero as initial coefficients.
the number of folds for cross-validation, default is 5.
a list with components:
the optimal pair of lambda1 and lambda2.
the misclassification rate of the optimal pair of lambda1 and lambda2.
a matrix of the misclassification rates for all pairs of lambdas tested.
Ren, J., He, T., Li, Y., Liu, S., Du, Y., Jiang, Y., Wu, C. (2017). Network-based regularization for high dimensional SNP data in the case <U+2013> control study of Type 2 diabetes. BMC Genetics.