Formatting date labels on bar plot

Question

My question

Suppose I have a Series like this.

In[10]: month_series
Out[10]: 
2016-01-01    4880
2016-02-01    4579
2016-03-01    6726
2016-04-01    1778
2016-05-01    3303
2016-06-01    5402
2016-07-01    1207
2016-08-01    6176
2016-09-01    5586
2016-10-01    2802
2016-11-01    6944
2016-12-01    3580
2017-01-01    9336
dtype: int64

I want to just construct a bar plot to compare month to month, which seems quite simple with

In[11]: month_series.plot(kind='bar')
Out[11]: 

I really don't need anything much more than that, but my issue is that the dates look terrible on the x-axis - I subsequently wanted to format the dates so that I just provide the year and month, some format like %Y-%m. How can I do so?

---

My struggles

So, looking at the documentation for pandas.Series.plot I see the xticks argument that can be passed, and figure I can use strftime to format the dates and pass as a sequence.. but this doesn't work because the ticks require numerical values or dates, not strings.

So then I figure I should just use raw matplotlib instead, so that I can use set_major_formatter with a DateFormatter to modify the tick labels. But if I just use plt.bar, that introduces a new issue - the entire date range is used, which makes sense.

In[17]: plt.bar(month_sereis.index, month_sereis.values)
Out[17]: <Container object of 13 artists>  

At this point I'm quite convinced I'm missing something and there's a trivially simple way to do this.

Answer 1

You can make use of matplotlib.axes.Axes.set_xticklabels whose argument labels accepts a string like sequence (anything printable with "%s" conversion) and then make use of DateTimeIndex.strftime to specify the format you'd want the labels to appear in.

ax = month_series.plot.bar()
# ax.set_xticklabels(month_series.index.to_period('M')) also works 
ax.set_xticklabels(month_series.index.strftime('%Y-%m'))
plt.show()