Radviz error from DataFrame which doesn't have sequantial index about yellowbrick HOT 3 OPEN

Hello @KimByoungmo and thank you for using Yellowbrick!

This is an issue that has been noted in the past (#1180), and we can definitely appreciate your use case. However, Yellowbrick is designed to work with or without pandas, treating X and y as numpy arrays. Because of that, we cannot simply change y[i] to y.iloc[i], as that would break Radviz's functionality for all users who are not passing in a dataframe.

That said, if you have any ideas about how to change the functionality so that Yellowbrick works more easily with both pandas and non-pandas data, we're all ears! Would you be interested in opening up a PR with some exploratory code to handle non-continuously indexed dataframes that passes both our pandas and our numpy integration tests?

from yellowbrick.

@rebeccabilbro

I think it would be better to use "isinstance" function
for instance, refer to below code

if isinstance(my_object,pd.Series) or isinstance(my_object,pd.DataFrame) :
##my_object.iloc[i]
else :
##my_object[i]

from yellowbrick.

@KimByoungmo looks like you're on the right track (though I think you'll also want an isinstance(my_object, numpy.array) in there too, right?)

We would welcome looking at a PR from you if you are open to contributing to Yellowbrick. We are a small team of unpaid volunteers, so we depend a lot on the contributions of our users to expand/change current functionality. You can check out our contributor's guide to see how to get started!

from yellowbrick.