from sklearn.metrics import auc
total = len(dat[target])
one_count = np.sum(dat[target])
one_count
33
zero_count = total - one_count
zero_count
766
lm = [y for _,y in sorted(zip(dat[i],dat[target]),reverse=True)]
x = np.arange(0,total+1)
y = np.append([0],np.cumsum(lm))
a = auc([0,total],[0,one_count])
aP = auc([0,one_count,total],[0,one_count,one_count]) - a
aR = auc(x,y) - a
fpr, tpr, thresholds = roc_curve(target, var)
auc_ = auc(fpr, tpr)
ar = auc_ * 2 - 1
print("Acc ratio:",aR/aP)
print("Acc ratio from ROC curve:",2*roc_auc_score(dat[target],dat[i])-1)
print("Acc ratio from ROC curve:",ar)
Acc ratio: 0.9278423925943509
Acc ratio from ROC curve: 0.775733839702508