cvsmac: A cross validation function for smac.

In smac: Sparse Multi-category Angle-Based Large-Margin Classifiers

Description

This function is used to perform cross validation based on a training data set to select the best tuning parameter.

Usage

cvsmac(x,y,kfold=5,lambda=NULL,nlambda=100,lambda.min=NULL,seed=0,weight=NULL,...)

Arguments

x	The usage of this argument is the same as that in the "smac" function.
y	The usage of this argument is the same as that in the "smac" function.
kfold	The number of subsamples for cross validation. This number should be less than the sample size of the class with the fewest obervations. A warning will be given if kfold is too large. Default is 5.
lambda	The usage of this argument is the same as that in the "smac" function.
nlambda	The usage of this argument is the same as that in the "smac" function.
lambda.min	The usage of this argument is the same as that in the "smac" function.
seed	The seed for generating the random split of the training data set. Default is 0.
weight	The usage of this argument is the same as that in the "smac" function.
...	Other arguments used by function "smac" that are not specified.

Value

lambda	The sequence of tuning parameters used in cross validation. Notice that the lambdas will be in a decreasing order.
beta0	The estimated intercepts with respect to each tuning parameter lambda.
beta	The estimated parameters of the predictors with respect to each tuning parameter lambda.
error	The total number of misclassifications with respect to each tuning parameter lambda. If weight is specified, each miscalssification is multiplied by its corresponding weight.
best.lambda	The sequence of lambda values that have the smallest cross validation error.
best.beta0	The intercepts that correspond to best.lambda.
best.beta	The parameters that correspond to best.lambda.
model	The fitted model with respect to each tuning parameter, using the entire training data set.
min.error	The minimum error in cross validation.

Author(s)

Chong Zhang, Guo Xian Yau and Yufeng Liu

References

C. Zhang and Y. Liu (2014). Multicategory Angle-based Large-margin Classification. Biometrika, 101(3), 625-640.

See Also

smac,predict.cvsmac

Examples

data(ex1.data)
cvsmac(ex1.data$ex1.x,ex1.data$ex1.y,loss="p",nlambda=30)

smac documentation built on May 1, 2019, 9:11 p.m.