ROCAUC function

[poppingcode]

Describe the issue
A clear and concise description of what the issue is.
when i use ROCAUC function,can i load a resnet-50 model?
for example :
model =resnet。。。。
viz = ROCAUC(model)

I want it to extend to other models

@DistrictDataLabs/team-oz-maintainers

[bbengfort]

@poppingcode if you're referring to a Keras or TensorFlow resnet-50 model, then I'll say that it's possible with a little bit of glue but still untested - although there is an open issue #397 to add contrib support for these types of estimators.

If you'd like to try to implement a non-scikit-learn model with yellowbrick, the key is to wrap it in an estimator, with approximately the following interface:

from sklearn.base import BaseEstimator 

class WrapperEstimator(BaseEstimator):
    
    def fit(self, X, y):
        return self

    def predict(self, X):
        return yhat 

    def score(self, X, y):
        return score 

Some models additionally need a predict_proba method; I'm not sure if ROC/AUC is one of those.

You can see an example of this for StatsModels in yellowbrick.contrib.statsmodels.

Does that help answer the question?