discreteControl: Define controlling parameters for discrete classifiers (NBLDA...

Description Usage Arguments Author(s) See Also Examples

View source: R/classify.R

Description

This function sets the control parameters for discrete classifiers (PLDA and NBLDA) while training the model.

Usage

1
2
3
4
discreteControl(method = "repeatedcv", number = 5, repeats = 10,
  rho = NULL, rhos = NULL, beta = 1, prior = NULL, alpha = NULL,
  truephi = NULL, foldIdx = NULL, tuneLength = 30,
  parallel = FALSE, ...)

Arguments

method

validation method. Support repeated cross validation only ("repeatedcv").

number

a positive integer. Number of folds.

repeats

a positive integer. Number of repeats.

rho

a single numeric value. This parameter is used as tuning parameter in PLDA classifier. It does not effect NBLDA classifier.

rhos

a numeric vector. If optimum parameter is searched among given values, this option shpould be used.

beta

parameter of Gamma distribution. See PLDA for details.

prior

prior probabilities of each class

alpha

a numeric value in the interval 0 and 1. It is used to apply power transformation through PLDA method.

truephi

a numeric value. If true value of genewise dispersion is known and constant for all genes, this parameter should be used.

foldIdx

a list including the fold indexes. Each element of this list is the vector indices of samples which are used as test set in this fold.

tuneLength

a positive integer. If there is a tuning parameter in the classifier, this value is used to define total number of tuning parameter to be searched.

parallel

if TRUE, parallel computing is performed.

...

further arguments. Deprecated.

Author(s)

Dincer Goksuluk, Gokmen Zararsiz, Selcuk Korkmaz, Vahap Eldem, Ahmet Ozturk and Ahmet Ergun Karaagaoglu

See Also

classify, trainControl, discreteControl

Examples

1
1L

dncR/MLSeq documentation built on May 17, 2020, 6:45 p.m.