sklearn如何计算二进制分类器的roc曲线下的面积？

Question

This might appear as a duplication of another question which has been asked here . However, I've looked at the answer there and still cannot understand how Scikit-learn calculates the area under the roc curve by testing only one threshold, which is the one provided in the:

y_pred = clf.predict(X_test) roc_auc_score(y_test, y_pred)

why it doesn't take multiple values (multiple y_test, y_pred that results from multiple thresholds)?any simplified explanation would be really appreciated.

Answer 1

The second argument for roc_auc_score() in this case should be the prediction probability obtained by clf.predict_proba(X_test) . The different thresholds are calculated inside this function on the basis of this prediction probabilities. There is an example for this in the documentation :

import numpy as np
from sklearn.metrics import roc_auc_score
y_true = np.array([0, 0, 1, 1])
y_scores = np.array([0.1, 0.4, 0.35, 0.8])
roc_auc_score(y_true, y_scores)
0.75

To understand how the roc_auc_score is caluclated it might be helpful to look at the roc_curve it self. This can be done with the function sklearn.metrics.roc_curve() . Example taken from the documentation :

import numpy as np
from sklearn import metrics
y = np.array([1, 1, 2, 2])
scores = np.array([0.1, 0.4, 0.35, 0.8])
fpr, tpr, thresholds = metrics.roc_curve(y, scores, pos_label=2)
fpr
array([ 0. ,  0.5,  0.5,  1. ])
tpr
array([ 0.5,  0.5,  1. ,  1. ])
thresholds
array([ 0.8 ,  0.4 ,  0.35,  0.1 ])

(Eventhough the y is different in the latter example it is still a binary classification with 2 being the positive class.)

As can be seen in the latter example the different thresholds are taken from the supplied scores .

The ROC Curve would be generated by putting the True Positive Rate tpr on the y-axis and the False Positive Rate fpr on the x-axis of a plot.