scikit Learn with Google app engine

Question

I am currently working on a data science project using python 2.7 framework which uses few modules like numpy,flask,pandas and have successfully deployed the application to Google cloud using app engine. I am now planning to use scikit-learn ML framework and my concern if it is compatible with google app engine?. I have seen some past forums where they mentioned sci-kit learn is not supported on App Engine. Wanted to check if this is still the case. Appreciate your input's here

Answer 1

You're referring to the Pure Python limitation in the Python standard environment sandbox.

One possible solution is using the GAE Python flexible environment, which doesn't have such limitation. While fundamentally it uses Google Compute Engine (GCE) VMs underneath it still retains some of the GAE convenience features which you'd need to handle yourself if going straight with a GCE solution.

See the Choosing an App Engine Environment guide.

Answer 2

It looks like sci-kit learn is not supported on App Engine, and the using managed VMs is the preferred solution. See the resolution of this bug.

Also see this similar question for another alternative approach.

I suggest you check out Google Compute Engine or Google container Engine.

Answer 3

2.7 Standard only allows you to import whitelisted/proprietary modules (listed in the Docs), whereas the Flex Environment supports all Python modules.

The new 2nd Generation Python 3.7 Standard Environment supports using all modules as well as many other features that 2.7 Standard doesn't.