[rbokeh] Hiding outlier points in boxplot

--
Steve Lianoglou
Bioinformatics Scientist
Cancer Immunology
Genentech

On Mon, Aug 15, 2016 at 12:47 PM, Steve Lianoglou <lianoglou.steve@gene.com> wrote:

Hi all,

Apologies if there's already a q&a or documentation about this, but I
can't find it, although I imagine this is quite common ...

Is there a way to go about hiding outlier points in a ly_boxplot? I'm
doing (the usual) drawing of a boxplot, then layering points over them
so that the user can interact (hover, etc.) the points in the ly_point
layer.

Something like this:

  figure(iris) %>%
    ly_boxplot(x=Species, y=Sepal.Length) %>%
    ly_points(x=catjitter(Species), y=Sepal.Length)

I want to remove the outlier circle outside the whiskers of the
boxplot that's drawn by `ly_boxplot` so that we only rely on
`ly_points` to draw the points (and avoid the double circle (there's
just one in this example, at the bottom of the "virginica" whisker)).

Playing with the alpha and color params for ly_boxplot will effect
both the circles and the box/whiskers, so I can't make my way around
it ...

If there isn't already a way to do this, and having a *very limited*
handle on the structure/design of the codebase, I feel like adding a
"with_outliers=TRUE" parameter to the `ly_boxplot`[0] function, and
using that to guard against running the `if (length(bp$out) > 0)[1]`
clause might be one way to deal with that ...

Still, I imagine I'm probably missing something here, so thanks in
advance for any help/pointers.

Cheers,
-steve

[0]: https://github.com/bokeh/rbokeh/blob/master/R/layer_stats.R#L254
[1]: https://github.com/bokeh/rbokeh/blob/master/R/layer_stats.R#L365

--
Steve Lianoglou
Bioinformatics Scientist
Cancer Immunology
Genentech

--
Steve Lianoglou
Bioinformatics Scientist
Cancer Immunology
Genentech