在Python中计算confusion_matrix

Question

I wan to calculate calculate True_Positive, False_Positive,False_Negative, True_Negative for three categories. I used to have two classes Cat Dog and this is the way i used to calculate my confusion_matrix

    Y_pred has either a cat or dog 
    y_true has either a cat or dog  
    confusion_matrix_output =confusion_matrix(y_true, y_pred) 
    True_Positive = confusion_matrix_output[0][0]
    False_Positive = confusion_matrix_output[0][1]
    False_Negative = confusion_matrix_output[1][0]
    True_Negative = confusion_matrix_output[1][1]

---

Now i have three classes 'Cat' 'Dog' 'rabbit'

Y_pred has Cat Dog rabbit
y_true has Cat Dog rabbit

How to calculate True_Positive, False_Positive,False_Negative, True_Negative ???

Answer 1

Now you have three classes, so it's not just positives and negatives anymore. You have to look at: Cat predicted as a cat, Dog predicted as Dog, Rabbit predicted as Rabbit, Dog predicted as cat, Cat predicted as Dog, and so on. You'll have 3 by 3 confusion matrix for this situation. Confusion matrix size is n by n, where n is the number of classes

sklearn.metrics.confusion_matrix abstracts away all that and creates an n by n matrix for you. Try this:

from sklearn.metrics import confusion_matrix
confusion_matrix_output =confusion_matrix(y_true, y_pred) 
Cat_P_Cat = confusion_matrix_output[0][0]