Barplot does not allow bar width to be set.

[sjoerdvansteenkiste]

The width argument can not be passed in barplot to specify the bar width. Since this is possible for boxplot I assume this was not intended.

[TomAugspurger]

Do you have an example?

In [26]: s = pd.Series([1, 2, 3])

In [27]: s.plot.bar(width=.1)
Out[27]: <matplotlib.axes._subplots.AxesSubplot at 0x1127b07f0>

In [28]: plt.figure()
Out[28]: <matplotlib.figure.Figure at 0x11a777390>

In [29]: s.plot.bar(width=.9)
Out[29]: <matplotlib.axes._subplots.AxesSubplot at 0x11a781128>

[sjoerdvansteenkiste]

tips = sns.load_dataset("tips")
ax = sns.barplot(x="day", y="total_bill", hue="sex", data=tips, width=.3)

If I look inside then I can not find a way to alter the internal width parameter through barplot, and hence the initial width of .8 is used, scaled according to the number of bars. And as you correctly pointed out bar does allow for its width to be altered.

[mwaskom]

It is intended.

[brevans]

This becomes a larger problem, especially on larger plots when hue is chosen. Please see:
http://stackoverflow.com/questions/36092363/seaborn-boxplots-changes-narrows-width-of-boxes-when-a-hue-is-chosen-how-migh

It forces you to choose between having tiny boxes (or violins, etc) or generating your own color palette and then having to create a legend from scratch.

[alexpetralia]

Was @brevans issue ever looked at? I'm having this same issue as well on barplots. If you put in the hue, you get thin bars.

[HannesHolste]

Agreed with @alexpetralia, this does seem to be unintended behavior when specifying hue and palette. Or if not unintended, then from a user's standpoint, the seaborn documentation around palette/hue is unclear or misleading.

I know of the solution suggested in: http://stackoverflow.com/questions/36092363/seaborn-boxplots-changes-narrows-width-of-boxes-when-a-hue-is-chosen-how-migh

I can submit a PR to clarify this behavior in the documentation if you'd like @mwaskom.