lift.chart: Lift Chart
In BioStaCs/BSCgas: Greedy AUC stepwise

Description Usage Arguments Details Value Note Author(s) References See Also Examples

Model evaluation: lift chart

1	lift.chart(prediction, labels, dec.point = 0.5, nclass = 10, colorize = T)

`prediction`	predicted result (double)
`labels`	real target lable (binary)
`dec.point`	cut off point, default=.5
`nclass`	Classifed in to n class, default=10
`colorize`	whether use color, default=F

Lift charts

Picture

Lift Charts

Yifan Yang

http://sweb.uky.edu/~yya234

AUC

##---- Should be DIRECTLY executable !! ----
##-- ==>  Define data, use random,
##--	or do  help(data=index)  for the standard data sets.

## The function is currently defined as
function (prediction, labels, dec.point = 0.5, nclass = 10, colorize = T) 
{
    if (nclass < 5) {
        print("nclass should be larger than (or equals) 5")
        nclass = 5
    }
    n <- length(labels)
    if (length(prediction) != n) {
        print("lengths of prediction and labels are different.")
        return(F)
    }
    if (n < nclass) {
        print("length of prediction should be larger than N-class.")
        return(F)
    }
    if (n%%nclass == 0) {
        col <- rainbow(nclass)
        n.perclass <- n/nclass
        pred.rate <- rep(0, nclass)
        tmp <- as.integer(sort(prediction, decreasing = T) > 
            dec.point)
        for (i in 1:nclass) {
            pred.rate[]
        }
    }
    else {
        col <- rainbow(nclass) + 1
        print("Extra 1 new class")
        n.perclass <- as.integer(n/nclass)
        n.res <- n%%nclass
        pred.rate <- rep(0, nclass)
    }
    return(pred.rate)
  }