add feature to change x-axis order for boxplots

[Tobias-Thomas]

Fixes #409, by adding by as a new optional parameter to boxplots. This by is the same, one would use for sort. It's default parameter is the identity.
Before merge, can someone please tell me, where I can add the documentation for this new parameter?
Thanks in advance.

[sethaxen]

@Tobias-Thomas thank you for the contribution, and I'm very sorry for the late review! I know I recommended the keyword by, but in retrospect, I think that's too generic of a name. What about something like sort_labels_by, or perhaps you have a better suggestion?

[Tobias-Thomas]

Oh, no problem. The only other idea, I would have is sort_x_by, since the parameter it sorts is called x.
What do you think?

[indymnv]

I was searching for a way to sort boxplots, but still is not solved or I am wrong?, is there a way I can help to fix this?

[Tobias-Thomas]

The fix I implemented was essentially quite easy, but as you can see above, we were still unsure about the name of this parameter, and I think we need a JuliaPlots member to decide on that.

[indymnv]

What about order to keep a similar idea like seaborn, sort_labels_by also can be OK, sort_x_by can be a little bit confused if later there is an implementation using horizontal boxplots.

[Tobias-Thomas]

Yea, I understand that point. Out of those options, I would prefer sort_labels_by.
I will do a commit in my branch (switching to this name) later this day and maybe we could ping Seth again, to see if he approves.

[codecov-commenter]

Codecov Report

Patch and project coverage have no change.

Comparison is base (90b5f4f) 26.52% compared to head (3164468) 26.52%.

Additional details and impacted files

@@           Coverage Diff           @@
##           master     #483   +/-   ##
=======================================
  Coverage   26.52%   26.52%           
=======================================
  Files          20       20           
  Lines        1248     1248           
=======================================
  Hits          331      331           
  Misses        917      917           

Impacted Files	Coverage Δ
src/boxplot.jl	0.00% <0.00%> (ø)

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.

[Tobias-Thomas]

@sethaxen, I updated the name and merged it into the current master. Is there anything left to do for me in this PR?

[jonathan-durbin]

Hi, any updates on this? I'm running into this issue while trying to figure out how to sort the x axis in a boxplot.

[Tobias-Thomas]

I think the fix would be ready for merge, but I do not know who to ping, to review the PR..

[indymnv]

Maybe @sethaxen , can help us with this PR

[sethaxen]

Sorry @Tobias-Thomas for the delay! Everything looks good to me! Thanks!

[Vaeryus]

So how do I use sort_labels_by now? Can I just put in an order for the x axis like:
sort_labels_by= [ "B" "C" "A"] or how would I use it?

[Tobias-Thomas]

So how do I use sort_labels_by now? Can I just put in an order for the x axis like: sort_labels_by= [ "B" "C" "A"] or how would I use it?

No, the function uses the same interface as Base.sort. That means that values can be sorted by a transformation of their values. So you need to write a function, which transforms the values into a space where they are sorted the way you want it.

[artkel]

Can someone please explain how the category sorting for boxplots now works? It is not quite clear to me from the previous comment. Would be nice to see an example. Thanks!

[artkel]

Can someone please explain how the category sorting for boxplots now works? It is not quite clear to me from the previous comment. Would be nice to see an example. Thanks!

Ok, I think I got it. The example below explains the logic. It worked for me.

# Define a function to map the categories to an order
function order_label(label::String)
    order_dict = Dict("B" => 1, "C" => 2, "A" => 3, "D" => 4)
    return get(order_dict, label, 0)  # Return 0 for any unexpected labels
end

# Now use this function in the sort_labels_by parameter
@df dataset_name boxplot(
    string.(:CategorcalVar),
    :NumericalVar,
    outliers=true,
    sort_labels_by=order_label,
    fillalpha=0.75,
    linewidth=1,
    legend=false
)

It would be nice to have a similar parameter for the violin plot.

[mkborregaard]

Before this PR the way to do that was to use levels! on the CategoricalArray x vector to define the sorting order. Is this really an improvemet on that?