When is a question considered a Data Science question and when is considered a coding question more appropriate for StackOverflow?

The title is self explanatory: where to draw the line between data science specific questions, suited for the Data Science SE, and coding questions that are more appropriate for StackOverflow?

For example, consider the question "How to use Pandas in conjunction with scikit-learn?". Which category should this belong to?

I think that the following simple logic should be applied. If a question is in its essence more about data science (models, approaches, methods, tools, etc.), where a programming language represents just a context, then such question fits Data Science SE site better. If, alternatively, the situation is reversed, where programming (coding) aspects are the core of the question, while data science represents context, such question should be asked on (or migrated to) StackOverflow (with a corresponding tag, i.e. r, python). I think that purely a fact of using a language, popular in data science community, as a context, does not justify a question to be considered on-topic on Data Science SE site. We can find an example of such situation in this purely R programming question.