mbinest {mt}

Binary Classification by Multiple Classifier

Description

Binary classification by multiple classifier.

Usage

mbinest(dat, cl, choices = NULL, method, pars=valipars(),...) 

Arguments

dat

A matrix or data frame containing the explanatory variables.

cl

A factor specifying the class for each observation.

choices

The vector or list of class labels to be chosen for binary classification. For details, see dat.sel.

method

Multiple classification methods to be used. For details, see maccest.

pars

A list of parameters of the resampling method. See valipars for details.

...

Additional parameters to method.

Value

A list with components:

all

All results of classification.

com

A matrix of the combinations of the binary class labels.

acc

A table of classification accuracy for the binary combination.

mar

Prediction margin.

auc

The area under receiver operating curve (AUC).

method

Classification methods used.

niter

Number of iterations.

sampling

Sampling scheme used.

nreps

Number of replications in each iteration if sampling is not loocv.

Author(s)

Wanchang Lin

See Also

maccest, maccest,valipars, dat.sel

Examples

## iris data set
data(iris)
dat    <- subset(iris, select = -Species)
cl     <- iris$Species
method <- c("svm","pcalda")

pars  <- valipars(sampling="cv",niter = 10, nreps = 5)
res   <- mbinest(dat,cl,choices=c("setosa"), method=method,
                  pars = pars, kernel="linear")

## combine prediction accuracy, AUC and margin 
z      <- round(cbind(res$acc,res$auc,res$mar),digits=3)
colnames(z) <- c(paste(method,".acc", sep=""),paste(method,".auc", sep=""),
                 paste(method,".mar", sep=""))