Fixes

[patrickfletcher]

Hi Bastian!

I really like your object oriented violin plots! I am inspired to rewrite several of my own plotting functions following this pattern. I found a few little things that I thought could be fixed (see commit messages).

A few more thoughts for improvements, which I plan to complete in my fork if you're interested:

• make a separate "MiniBoxPlot" class to simplify the Violin code. Violin will then contain one of these. This will be used via a plot function (like violinplot.m) as a simple standalone object-oriented alternative to Matlab's boxplot function (in plotstyle='compact' mode).
• in violinplot, support plotting a matrix of violins, when rows (observations) are grouped according to a grouping variable, and columns represent variables (with names optionally given).

Best,

Patrick

[bastibe]

Cool! I like it!

Just one question: In bd5059a, did you change the default box width from 0.01 to 0.03? Could you post a screen shot of the two versions?

As for factoring out a MiniBoxPlot, I like the idea in general, but in my personal experience, more generality is not always to the benefit of the code. We would have to decide with an actual code example if the generalization is worth the tradeoff in readability.

Regarding a matrix of violin option, I am certainly open to that, so long as it doesn't break or change existing functionality.

[patrickfletcher]

RE: box width. I updated things to be exactly as in your original version (new vs original images attached). Also I thought the default violin width (0.3) could be encoded in the input parser (see new commit).

As for refactoring, I'm not sure yet whether it'll help or hinder. For that idea and also for the matrix violin options, I'll propose those as a separate issue (for discussion) or pull requests (for the matrix) when I get more done.

[bastibe]

Thank you so very much! I love it!

[patrickfletcher]

Glad to help. Thank you for this project :)