ValueError: multilabel-indicator is not supported for confusion matrix

问题

I am encountering this error when trying to use the confusion matrix in my binary classification problem. Y and Yhat are both numpy arrays. I have tried the .argmax as the proposed solution - I don't get the error anymore but the output is not the confusion matrix that I know.

Accuracy: 0.9982449999999999
Accuracy: 0.9983374013937532

shape of y =  (1, 200000)
shape of yhat =  (1, 200000)

The error

 ValueError                                Traceback (most recent call last) <ipython-input-13-ebb660b4585a> in <module>()
     12 print("shape of yhat = ", yhat.shape)
     13 
---> 14 cm = confusion_matrix(y,yhat)
     15 
     16 print("confusion matrix = ", cm)

/anaconda3/lib/python3.6/site-packages/sklearn/metrics/classification.py in confusion_matrix(y_true, y_pred, labels, sample_weight)
    250     y_type, y_true, y_pred = _check_targets(y_true, y_pred)
    251     if y_type not in ("binary", "multiclass"):
--> 252         raise ValueError("%s is not supported" % y_type)
    253 
    254     if labels is None:

ValueError: multilabel-indicator is not supported

回答1:

Append your code as follows

cm = confusion_matrix(y.argmax(axis=1),yhat.argmax(axis=1))

Asked here before

来源：https://stackoverflow.com/questions/52222515/valueerror-multilabel-indicator-is-not-supported-for-confusion-matrix