PairPlots.jl
PairPlots.jl copied to clipboard
sefffal
Add support for categorical columns
It would be nice to have some support for categorical columns (e.g. columns of strings).
We are building a stack on top of DataScienceTraits.jl to recognize the different scientific types of columns and dispatch different behavior accordingly: https://github.com/JuliaML/DataScienceTraits.jl
Would it be a good idea to add the package as a dependency, and then adjust the plots whenever a column is Categorical?
Notice that the package is lightweight, and all specializations live in package extensions.
Interesting. Currently the visualizations layers are chosen based on their position (diagonal or body) but I could see this being extended to different visualizations per pair of variables. It might need some restructuring.
How would you expect categorical plots to look?
I would expect categorical variables to show as bar plots in the diagonal and some other custom visualization in the off-diagonal plots. We could also imagine a situation where users just want to use categorical columns to color the plots of continuous variables.
@sefffal do you have any idea of how much work is required to refactor PairPlots.jl to include such features? It would be amazing to use the package in more advanced data science contexts.
We are using the package in university short-courses, and students always love it.
Appreciate the bump @juliohm . I am still interested in adding this feature but it's hard for me to justify prioritizing it vs my own research at the present moment.
Sorry for the disappointing answer but it's better to set realistic expectations.
If you have eg a student willing to work on this I could support them in making the necessary changes.
Thank you for considering the feature, we use PairPlots.jl all the time in our explorations. No need to rush, I was just curious about your perspective on the issue. :)
Em sáb., 27 de abr. de 2024, 13:13, William Thompson < @.***> escreveu:
Appreciate the bump @juliohm https://github.com/juliohm . I am still interested in adding this feature but it's hard for me to justify prioritizing it vs my own research at the present moment.
Sorry for the disappointing answer but it's better to set realistic expectations.
If you have eg a student willing to work on this I could support them in making the necessary changes.
— Reply to this email directly, view it on GitHub https://github.com/sefffal/PairPlots.jl/issues/33#issuecomment-2080993783, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAZQW3LHC7LD7YTQYSYVF4LY7PFC3AVCNFSM6AAAAABABYELBGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDAOBQHE4TGNZYGM . You are receiving this because you were mentioned.Message ID: @.***>