MarginOrderedPruning.Bagging: MarginOrderedPruning.Bagging

In adabag: Applies Multiclass AdaBoost.M1, SAMME and Bagging

MarginOrderedPruning.Bagging

Description

Margin-based ordered aggregation for bagging pruning

Usage

MarginOrderedPruning.Bagging(baggingObject, trainingset, pruningset, 
	marginType = "unsupervised", doTrace = TRUE)

Arguments

baggingObject	fitted model object of class bagging
trainingset	the training set of the bagging object
pruningset	a set aside dataset for bagging pruning
marginType	if "unsupervised" (by default) the margin is the difference between the proportions of votes of the first and second most popular classes. Else the margin is calculated as the difference between the proportion of votes of the correct class and the most popular among the other classes
doTrace	If set to TRUE, give a more verbose output as MarginOrderedPruning.Bagging is running

Value

Returns a list with the following components:

prunedBagging	a pruned bagging object
AccuracyOrderedEnsemblePruningSet	Accuracy of each ordered ensemble on pruning set

Note

Questions about this function should be sent to Li Guo

Author(s)

Li Guo [e-mail]

References

Guo, L. and Boukir, S. (2013): "Margin-based ordered aggregation for ensemble pruning". Pattern Recognition Letters, 34(6), 603-609.

See Also

bagging, predict.bagging

Examples

## mlbench package should be loaded
library(mlbench)
data(Satellite)
## Separate data into 3 parts: training set, pruning set and test set
ind <- sample(3, nrow(Satellite), replace = TRUE, prob=c(0.3, 0.2,0.5))

## create bagging with training set
#increase mfinal in your own execution of this example to see 
#the real usefulness of this function
Satellite.bagging<-bagging(classes~.,data=Satellite[ind==1,],mfinal=3)
#Satellite.bagging.pred<-predict(Satellite.bagging,Satellite[ind==3,])

##pruning bagging
Satellite.bagging.pruning<-MarginOrderedPruning.Bagging(Satellite.bagging,
Satellite[ind==1,],Satellite[ind==2,])
#Satellite.bagging.pruning.pred<-predict(Satellite.bagging.pruning$prunedBagging,
#Satellite[ind==3,])

## create bagging with training and pruning set
#This example has been hidden to fulfill execution time <5s 
#Satellite.bagging2<-bagging(classes~.,data=Satellite[ind!=3,],25)
#Satellite.bagging2.pred<-predict(Satellite.bagging2,Satellite[ind==3,])

adabag documentation built on May 31, 2023, 9:09 p.m.