adabag (version 2.1)

margins: Calculates the margins

Description

Calculates the margins of an adaboost.M1 or bagging classifier for a data frame

Usage

margins(object, newdata)

Arguments

object

This object must be the output of one of the functions bagging, adaboost.M1, predict.bagging or predict.boosting. This is assumed to be the result of some function that produces an object with two comp

newdata

The same data frame used for building the object

Value

• An object of class margins, which is a list with only a component:
• margins
  a vector with the margins.

Details

Intuitively, the margin for an observation is related to the certainty of its classification. It is calculated as the difference between the support of the correct class and the maximum support of an incorrect class

References

Alfaro, E., Gamez, M. and Garcia, N. (2007): ``A Boosting approach for corporate failure prediction''. Applied Intelligence. 27, pp. 29--37. Schapire, R.E., Freund, Y., Bartlett, P. and Lee, W.S. (1998): ``Boosting the margin: A new explanation for the effectiveness of voting methods''. The Annals of Statistics, vol 26, 5, pp. 1651--1686.

See Also

adaboost.M1, predict.boosting, bagging, predict.bagging,

Examples

#Iris example
library(rpart)
data(iris)
names(iris)<-c("LS","AS","LP","AP","Especies")
sub <- c(sample(1:50, 25), sample(51:100, 25), sample(101:150, 25))
iris.adaboost <- adaboost.M1(Especies ~ ., data=iris[sub,], mfinal=10)
margins(iris.adaboost,iris[sub,])->iris.margins # training set

# test set
iris.predboosting<- predict.boosting(iris.adaboost, newdata=iris[-sub,])
margins(iris.predboosting,iris[-sub,])->iris.predmargins 


#Examples with bagging
iris.bagging <- bagging(Especies ~ ., data=iris[sub,], mfinal=10)
margins(iris.bagging,iris[sub,])->iris.baggging.margins # training set


iris.predbagging<- predict.bagging(iris.bagging, newdata=iris[-sub,])
margins(iris.predbagging,iris[-sub,])->iris.baggging.predmargins # test set

# We change the name to the margins object to save space 
# and use the same code for the graph
#iris.margins<-iris.predmargins 
#iris.margins<-iris.baggging.margins
#iris.margins<-iris.baggging.predmargins

# Example of a graph using the margins
par(bg="lightyellow")

# 1 Rotulos en castellano, con disculpas por la falta de acentos por el codigo
plot(sort(iris.margins[[1]]), (1:length(iris.margins[[1]]))/length(iris.margins[[1]]),
type="l", xlim=c(-1,1),main="Grafico de la distribucion de los margenes", xlab="m", 
ylab="% conjunto entrenamiento", col="blue3")
abline(v=0, col="red",lty=2)

# 2 English headings
plot(sort(iris.margins[[1]]), (1:length(iris.margins[[1]]))/length(iris.margins[[1]]), 
type="l", xlim=c(-1,1),main="Margin cumulative distribution graph", xlab="m", 
ylab="% training set", col="blue3")
abline(v=0, col="red",lty=2)


## rpart and mlbench libraries should be loaded
library(rpart)
library(mlbench)
data(BreastCancer)
l <- length(BreastCancer[,1])
sub <- sample(1:l,2*l/3)

BC.adaboost <- adaboost.M1(Class ~.,data=BreastCancer[sub,-1],mfinal=25, control=rpart.control(maxdepth=3))
BC.adaboost.pred <- predict.boosting(BC.adaboost,newdata=BreastCancer[-sub,-1])

BC.margins<-margins(BC.adaboost,BreastCancer[sub,-1]) # training set
BC.predmargins<-margins(BC.adaboost.pred,BreastCancer[-sub,-1]) # test set