I didn't know this is not "okay" to do. I'm agnostic about it I guess. But also don't mind reducing LGTM warnings.

I would leave this open for another member to comment at least :)

I’m surprised this is a LGTM warning. I am +0 on this change.

I’m surprised this is a LGTM warning. I am +0 on this change.

Yeah, me too. I wonder what it's trying to avoid.

From LGTM https://lgtm.com/rules/1818040193/
Importing a module twice using the import xxx and from xxx import yyy is confusing.

also +0

IMO that LGTM.org warning is a bit controversial. Just because one does

import sklearn

somewhere (for instance, to check version), it would require people to use the full import paths everywhere e.g. sklearn.linear_model.LogisticRegression. I don't think that it's realistic and that we should enforce it.

Closing as it doesn't look like we are going get a +2 vote on this one. Sorry, @Kibugu and thanks for contributing. There are likely other warnings in need of fixing :)

Maybe @Kibugu would like to investigate if there's a way to configure our use of lgtm.com to disregard this kind of alert.

Hint: https://lgtm.com/help/lgtm/lgtm.yml-configuration-file in section "queries"