I request to have a new geom, geom_violin2, in which the drawn quantiles match the data rather than the density estimate.

[R180]

Following-up on the issue described at #2088 I request to have a new geom--it could be called geom_violin2 or perhaps geom_viola--in which the drawn quantiles match the data rather than being derived from the density estimate.

This will allow those of us who have an aesthetic or data-analytic preference for data-derived rather than density-estimated medians (for example) to be represented in the violin. Moreover, it would not introduce a breaking change because it would be a completely new geom that would co-exist with geom_violin.

[teunbrand]

Thank you for the suggestion. The upcoming version adapts geom_violin() to use data quantiles rather than density based quantiles, see #4120.