ensembleRfCV: Estimate classification performance using cross-validation...

In singha53/amritr: Accessory Functions Used by Amrit Singh

Description

Estimate classification performance using cross-validation using an random forest classifier

Usage

ensembleRfCV(X.trainList, y.train, panelLength = 15, filter = "none",
  topranked = 50, keepVarList = NULL, M = 5, folds = 5,
  progressBar = FALSE)

Arguments

X.trainList	list of training datasets (nxpi); i number of elements
y.train	n-vector of class labels (must be a factor)
filter	pre-filtering of initial datasets - "none" or "p.value"
topranked	Number of topranked features based on differential expression to use to build classifer
keepVarList	which variables to keep and not omit (set to NULL if no variables are forced to be kept)
M	# of folds
folds	list of length M, where each element contains the indices for samples for a given fold
progressBar	(TRUE/FALSE) - show progress bar or not
alphaList	list of alpha values
lambdaList	list of lambda values
family	can be "binomial" or "multinomial"

Value

error computes error rate (each group, overall and balanced error rate)

perfTest classification performance measures

singha53/amritr documentation built on July 21, 2019, 3:46 p.m.