Suggestion, heatmap to show pairwise correlations #feature #new_plot

[ewallace]

Hi bayesplot team, Thank you for making a brilliant and very usable package. The mcmc_scatter plot is very useful for visualizing the correlations between two known parameters. However it is not useful at visualizing which parameters are correlated in general. mcmc_pairs will do this in detail, but is hard to implement and prohibitively slow for more than a dozen or so paramters.

I suggest a heatmap of correlations: first, calculate pairwise correlations between (a subset of) parameters; second, plot a heatmap using geom_tile with the correlations as fill. That would be a quick visual cue to what is correlated with what.

Please accept my apologies if this was already discussed. Edward

[jgabry]

@ewallace Sorry for the slow response. It's been a busy few weeks. Thanks for the suggestion! I don't think this has been proposed for the package before. Any interest in trying it out and reporting back? A pull request is always welcome of course, but even just a simple single example is great and we can take it from there.

[ewallace]

@jgabry I will give that a try!

[jgabry]

Great, thanks!

On Tue, Dec 19, 2017 at 3:26 PM Edward Wallace [email protected] wrote:

@jgabry https://github.com/jgabry I will give that a try!

— You are receiving this because you were mentioned.

Reply to this email directly, view it on GitHub https://github.com/stan-dev/bayesplot/issues/130#issuecomment-352876609, or mute the thread https://github.com/notifications/unsubscribe-auth/AHb4Qxc01iqoLbLGp7JDHHZNM8t6Y-Llks5tCBwLgaJpZM4Q061B .

[jgabry]

@ewallace I’m circling back through old issues and I saw this one. Any interested in contributing this? Happy to help.

_{Sent with GitHawk}