bacP <-function(trueLabels,predictedLabels){
	#just check if both input arguments are of type factor
	if(!is.factor(trueLabels) || ! is.factor(predictedLabels)){
		throw("both arguments have to be a factor")
	}
	#what classes are possbible
	posClasses <- levels(trueLabels)
	#how many different classes?
	numClass <- length(posClasses)
	#preallocate space
	accWClasses <- rep(0,numClass)
	#loop through different classes and calculate within class accuracy
	for (i in 1:numClass){
		allC <- which(trueLabels==posClasses[i])
		hits <- which(predictedLabels[allC] == posClasses[i])
		accWClasses[i] <- length(hits) / length(allC)
	}
	return(mean(accWClasses))
}