h2oai / mli-resources Goto Github PK

• Shouldn't be an issue, but just to double check.

In order to successfully install examples using Docker I did the following changes:

• There seems to be missing step which clones mli-resources GitHub repository. Perhaps RUN git clone https://github.com/h2oai/mli-resources.git should be added to Dockerfile (I cloned repo manually).
• Jupyter refuses to start under root - consider adding --allow-root parameter: docker run -i -t -p 8888:8888 <image_id> /bin/bash -c "/opt/conda/bin/conda install jupyter -y --quiet && /opt/conda/bin/jupyter notebook --notebook-dir=/mli-resources --ip='*' --port=8888 --no-browser --allow-root" and/or suggest to run it under a normal user.
• Perhaps it would be worth to suggest user to use URL provided by Jupyter after it starts - it contains auth token (README.md, step 9.).

In given example a GBM is used to create tree. but TREPAN generates sample to satisfy minimum number of samples required(Smin). how does GBM do that here?

Following the instruction to build the Docker image,

3. Fetch the Dockerfile from the mli-resources repo.
   $ curl https://raw.githubusercontent.com/h2oai/mli-resources/master/anaconda_py36_h2o_xgboost_graphviz/Dockerfile > anaconda_py36_h2o_xgboost_graphviz/Dockerfile

The resulted Docker file contains the status code rather the content itself. Do this instead,
curl https://raw.githubusercontent.com/h2oai/mli-resources/master/anaconda_py36_h2o_xgboost_graphviz/Dockerfile | Select-Object -ExpandProperty Content

• Would be good to have necessary Python packages set up in a requirements.txt

Thanks everyone for this amazing source of model interpretation. I have a question in LOCO, regarding how H2O GBM deals with missing value.

In the markdown, it is said that H2O GBM deals with missing value by following the majority path in the tree. However in the H2O document, I think it means that H2O GBM deals with missing value by treating it as a new category and minimize the loss function. And in testing it will follow the missing value path optimized in training.

Do I have some missunderstanding here?

Thanks,
Sandy