Feature_names is redundant when using jsonData in predict. Also meta is redundant in response, as meta can be combined in the jsonData.

[chennupushkar]

When using jsonData, why does the predict signature still require features_names as an argument in the Predict function?

        """
        Return a prediction.

        Parameters
        ----------
        X : array-like
        feature_names : array of feature names (optional)
        """
        print("Predict called - will run identity function")
        return X```
. Also the response of the Predict function when using jsonData has meta:{} in it. When jsonData is suppose to be generic json why is predict function returning meta?
```curl -g http://localhost:5000/predict --data-urlencode 'json={"jsonData": {"ndarray": [[4.098, 2.128, 9.715, 2.977]]}}'
{"jsonData":{"my_predictions":[[3.9589564268624405e-18,4.098772877346702e-09,0.9999999959012271]]},"meta":{}}```

[ukclivecox]

Meta data is for the whole request and is understood by the engine component that manages the requests. So makes sense to keep it as a core Seldon feature for meta irrespective of data payload.

I agree feature names could be made a keyword argument which defaults to None.

seldon-core/python/seldon_core/user_model.py

Line 26 in 559f897

def predict(self, X: np.ndarray, names: Iterable[str], meta: Dict = None) -> Union[

[ukclivecox]

Closing. Please open if issue.