How to access the minimal dependencies of scikit-learn

[mfeurer]

Scikit-learn 0.18.1 requires numpy>=1.6.1 ad scipy>=0.9. This seems to be pretty constant during the latest versions of scikit-learn. In case this is about to change, can this be queried dynamically @amueller? Otherwise, it might be good have a dictionary, mapping from scikit-learn version to the dependency versions.

[amueller]

Sorry, didn't see the ping. We just upgraded. setup.py has
SCIPY_MIN_VERSION = '0.13.3'
NUMPY_MIN_VERSION = '1.8.2'

[mfeurer]

Thanks for letting us know. setup.py contains the relevant information, but we don't have access to that at runtime, so probably a dictionary would be best, mapping from scikit-learn version to dependencies?

[PGijsbers]

For reference: the reason we want this is to specify versions required when serializing a flow, so we have access to this information to deserialize it later.