This function calculates a Gini coefficient based on a Receiver Operator Curve.
Arguments
- pred
Logit/scores/probabilities to be compared against actuals
- act
This should be a column containing outcomes in a boolean form either as a factor or number
See also
AUC roc giniChart
Other creditrisk:
giniChart(),
logit.odd(),
logit.prob(),
odd.logit(),
odd.prob(),
prob.logit(),
prob.odd(),
scaledScore()
Examples
sampledata<- data.frame(val= rnorm(100) , outcome=rbinom(100,1,.8))
giniCoef(sampledata$val,sampledata$outcome)
#> [1] 0.21125