Dear R-helper, I am trying to compare two AUCs derived from same population, but one is derived by adding extra features on top of the other. I found the codes below from R-help. I would like to know, in my case, whether I could use cROC, and also how can I got the right correlation (r)? I have been back to the original paper(Hanley 1983). It suggested to calculate rN(correlation coefficient for the discriminant value for +1 labeled value) and rA(correlation coefficient for the discriminant value for -1 labeled value) and then calculate the average correlation and look up the Table they provided (row: average correlation, column: average AUC). However, in that Table the Average AUC need to be bigger than 0.70, which limits the search (always need to have average AUC larger than 0.70). Would you please suggested some way to calculate the correlation ? Thank you very much! Cheers, Xiao From: Bernardo Rangel Tura> Date: Thu 16 Dec 2004 - 07:30:37 EST > > seROC<-function(AUC,na,nn){ > a<-AUC > q1<-a/(2-a) > q2<-(2*a^2)/(1+a) > se<-sqrt((a*(1-a)+(na-1)*(q1-a^2)+(nn-1)*(q2-a^2))/(nn*na)) > se > } > > cROC<-function(AUC1,na1,nn1,AUC2,na2,nn2,r){ > se1<-seROC(AUC1,na1,nn1) > se2<-seROC(AUC2,na2,nn2) > > sed<-sqrt(se1^2+se2^2-2*r*se1*se2) > zad<-(AUC1-AUC2)/sed > p<-dnorm(zad) > a<-list(zad,p) > a > }
> Dear R-helper, > > I am trying to compare two AUCs derived from same population, but one > is derived by adding extra features on top of the other. I found the > codes below from R-help. I would like to know, in my case, whether I > could use cROC, and also how can I got the right correlation (r)? > I have been back to the original paper(Hanley 1983). It suggested to > calculate rN(correlation coefficient for the discriminant value for +1 > labeled value) and rA(correlation coefficient for the discriminant > value for -1 labeled value) and then calculate the average correlation > and look up the Table they provided (row: average correlation, column: > average AUC). However, in that Table the Average AUC need to be bigger > than 0.70, which limits the search (always need to have average AUC > larger than 0.70). Would you please suggested some way to calculate > the correlation ? > > > Thank you very much! > > Cheers, > > Xiao > > > > > From: Bernardo Rangel Tura >> Date: Thu 16 Dec 2004 - 07:30:37 EST >> >> seROC<-function(AUC,na,nn){ >> a<-AUC >> q1<-a/(2-a) >> q2<-(2*a^2)/(1+a) >> se<-sqrt((a*(1-a)+(na-1)*(q1-a^2)+(nn-1)*(q2-a^2))/(nn*na)) >> se >> } >> >> cROC<-function(AUC1,na1,nn1,AUC2,na2,nn2,r){ >> se1<-seROC(AUC1,na1,nn1) >> se2<-seROC(AUC2,na2,nn2) >> >> sed<-sqrt(se1^2+se2^2-2*r*se1*se2) >> zad<-(AUC1-AUC2)/sed >> p<-dnorm(zad) >> a<-list(zad,p) >> a >> } >