tune.it: tune.it

In aucm: AUC Maximization

Description

Tuning methods.

Usage

tune.it (formula, dat.train, dat.tune, method, kernel, verbose=TRUE, step.size = 2)
get.gamma0 (formula, dat.train)

Arguments

formula	a formula object.
dat.train	a data frame. Training data
dat.tune	a data frame. Tuning data. If NULL, gacv is done.
method	a string. "svm" or "rauc"
kernel	a string. "rbf" or "linear"
verbose	logical
step.size	step size for lambda as in step.size^seq(-3,3,1), please see code for more details

Value

tune.it returns a vector of one or two elements: lamcost and gamma, depending on the kernel.
get.gamma0 returns .5/quantile(dist., c(.9,.75,.5)), where dist. is the Eucledian distance between objects from two classes.

Author(s)

Shuxin Yin
Youyi Fong [e-mail]
Krisztian Sebestyen [e-mail]

aucm documentation built on Jan. 11, 2020, 9:43 a.m.