1
$\begingroup$

I produced a clustermap as part of an attempt to visualise any multicollinearity in a dataset and it's occurred to me that I actually don't really know how to interpret it.

There is plenty of documentation out there on how to build them (I used seaborn in python) but aside from the obvious, look at the colour scale to see how intensely correlated everything is, I'm not sure how one might put into words how to interpret this kind of graph. Especially in terms of explaining in a report of thesis type setting the usefulness of doing one and what specific results might mean.

I mean this question to be a general one where others can look for a basic guide on what kinds of features they can talk about and what they mean, but if anyone wants context from the one I produced:

enter image description here

$\endgroup$

1 Answer 1

1
$\begingroup$

If you have a lot of features isn't likely to have a easy interpretation out-of-the-box despite the reder and bluer ones have more positive and negative correlation and the white ones don't have as much. But in terms of features you could make a clustering from your clustermap, for example the positive correlated features, the negative ones and so on. But one need the final goal of your analysis to elaborate a better answer.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.