Floris Hoogenboom

Since this is a problem that is likely still encountered by many it may be good to write here that in the dev-0.20 version of sklearn [OneHotEncoder](http://scikit-learn.org/dev/modules/generated/sklearn.preprocessing.OneHotEncoder.html#sklearn.preprocessing.OneHotEncoder) directly supports categorical...

You are right about the first point! I've noticed that #7 introduced some other small bugs. I'll create a new PR to fix those. Regarding the PyPi package: I think...

Sure, [FlorisHoogenboom](https://pypi.org/user/FlorisHoogenboom/)

When using XGBoost 0.90 or up I would propose using the `XGBRanker` object that has been made available as part of the `scikit-learn` api there instead of relying on this...

You are completely right. This is fixxed (as a by product) in #12 I believe.

Thanks for flagging this. It shouldn't really be an issue since this `Lambda` layer is only used to create a flow though a subset of the network to get the...

Hi @dirkkul, I'm experiencing different behaviour though. Also judging from this section it seems that the current create does not function as an add or replace? https://github.com/semi-technologies/weaviate-python-client/blob/572a9d401eda68bc17a9f6e876f996ac6d1a9213/weaviate/data/crud_data.py#L139 Am I missing...