DOC: clarify boxplot whiskers extension

[neutrinoceros]

As I'm getting started with box plots, I found confusing that the DataFrame.boxplot docstring didn't mention that whiskers extend at most to 1.5 * IQR away from the box's edges.

• passes black pandas
• passes git diff upstream/master -u -- "*.py" | flake8 --diff
• whatsnew entry -> is it required for such a minor change ?

[MarcoGorelli]

Thanks @neutrinoceros , agreed that this clarifies things

[joooeey]

This doc change is sorely needed. Thanks for doing it! However, your explanation is wrong. The whiskers only ever extend to 1.5*IQR if there happens to be a data point exactly there. Here's how it actually works: https://stackoverflow.com/a/49781675/4691830 .

I suggest the following wording:

By default, they extend no more than `1.5 * IQR (IQR = Q3 - Q1)` from the edges of the box, ending at the farthest data point within that interval. Outliers are plotted as separate dots.

[neutrinoceros]

Thanks @joooeey for clarifying this, I applied your suggestion !

[MarcoGorelli]

However, your explanation is wrong

@joooeey are you sure it's wrong? When the OP wrote

they extend up to 1.5 * IQR (IQR = Q3 - Q1) from the edges of the box, or to the farthest
data point if it lies within this interval

I think they meant that the whiskers only extend there if there's a point there, which I think is what you're saying.

I suggest the following wording:

Having said that, thanks for this, I do think your wording is very clear

[MarcoGorelli]

@neutrinoceros did you push it? :) I don't (yet) see anything here

[neutrinoceros]

I think they meant that the whiskers only extend there if there's a point there, which I think is what you're saying.

I believe my own wording was correct if you read the "or" and the "within that interval" as an inclusive, which showed my own misunderstanding.

@neutrinoceros did you push it? :) I don't (yet) see anything here

Oops, you're right, the push failed for some reason.

[WillAyd]

Great thanks @neutrinoceros