如何从sklearn的gridsearchcv中获取灵敏度和特异性（真阳性率和真阴性率）？

Question

I've been using Gridsearchcv with a RBF SVM (binary classifier) to obtain validation accuracy heatmaps. The code I've used is pretty much straight from SKlearn's website. Is there a way to find the sensitivity and specificity from this? As in for the range of parameter values used by Gridsearchcv?

Answer 1

If your problem is binary or multi-class classification then confusion matrix might be what you're looking for.

from sklearn.metrics import confusion_matrix

y_true = [2, 0, 2, 2, 0, 1]
y_pred = [0, 0, 2, 2, 0, 2]
confusion_matrix(y_true, y_pred)

array([[2, 0, 0],
       [0, 0, 1],
       [1, 0, 2]])

The explanation is the following :

For the examples belonging to class 0, the estimator predicted 100% of them correctly (2/2).
For the examples belonging to class 1, the estimator was 100% wrong because it predicted the only example to class 2.
For the examples belonging to class 2, the estimator was 66% correct (2/3) because it predicted 2 examples to class 2 and 1 to class 0.

For binary classification :

y_true = [1, 0, 1, 0, 0, 1]
y_pred = [1, 0, 1, 1, 0, 1]

cm = confusion_matrix(y_true, y_pred)
print cm

tp = float(cm[0][0])/np.sum(cm[0])
tn = float(cm[1][1])/np.sum(cm[1])

print tp
print tn

[[2 1]
 [0 3]]
0.666666666667
1.0

About the parameters used in your GridSearchCV , you can found them in the grid_scores_ attribute.