Python : Different behaviour of DatetimeIndex while plotting line and bar plots using DataFrame

Question

I have a DataFrame with the row index as a DatetimeIndex.

This index is coming up differently on the x-axis while I am making line and bar plots. My code is as follows:

start_date = datetime.datetime.strptime('2017-02-20', '%Y-%m-%d').date()
end_date = datetime.datetime.strptime('2017-02-23', '%Y-%m-%d').date()

daterange = pd.date_range(start_date, end_date)
df = pd.DataFrame(index = daterange, data = {'Male':[12, 23, 13, 11], 'Female': [10, 25, 15, 9]})


df.plot(kind='line')
df.plot(kind='bar', stacked = False, grid=1)

The plots i am obtaining are as follows. Line plot with nice formatting of dates on x-axis:

Line plot with nice formatting of dates on x-axis

Bar plot without formatting of dates on x-axis:

Bar plot without formatting of dates on x-axis

In the line plot the x-axis labels are well formatted with the month and year on left corner and the dates used as x-ticks. But in the bar plot, the entire date along with the time (00:00:00) is shown unlike in the line plot.

How can I get the proper formatting of dates on x-axis in the bar plot and without the time being shown?

Sebastian Wozny · Accepted Answer · 2017-02-25 12:11:10Z

2

The problem is in the source code of pandas. You cannot get the bar plot to use the pd.DateTimeFormatter without deriving custom subclasses or using matplotlib directly.

In line 1766 (1784 in the dev version) of pandas.tools.plotting the datetime formatting for LinePlot is done. This is not present in BarPlot, for reasons that I can only hypothesize:

Line charts are intended to print timeseries data, whereas the same does not necessarily make sense for bar charts.

I would still like to see bar plots being able to format dates properly without using matplotlib, so you might want to open an issue with the pandas project.

With matplotlib directly:

import pandas as pd
import datetime
import matplotlib.ticker as ticker
import matplotlib.pyplot as plt

start_date = datetime.datetime.strptime('2017-02-20', '%Y-%m-%d').date()
end_date = datetime.datetime.strptime('2017-02-23', '%Y-%m-%d').date()
daterange = pd.date_range(start_date, end_date)
df = pd.DataFrame(index = daterange, data = {'Male':[12, 23, 13, 11], 'Female': [10, 25, 15, 9]})

ax=df.plot.bar(xticks=df.index.month, stacked = False, grid=1)
ticklabels = [item.strftime('%b %d') for item in df.index]
ax.xaxis.set_major_formatter(ticker.FixedFormatter(ticklabels))
plt.gcf().autofmt_xdate()

plt.show()

edited Feb 25, 2017 at 12:11

answered Feb 25, 2017 at 12:05

Sebastian Wozny

17.6k7 gold badges55 silver badges73 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

zingsy Over a year ago

With this solution I can use different directives of strftime() function to get weekday names and more for adding additional details to my graph like %a for getting the abbreviated weekday name. I was using the same code as your solution to get weekday names on the xticks for the line plot but it doesnt work with it giving me only an empty plot. Is it clashing with the already present formatting for the LinePlot ?

ImportanceOfBeingErnest · Accepted Answer · 2017-02-25 12:20:53Z

1

Bar plots are generally meant to be used to plot categorical data. That means that unlike in the line plot, the x-values are simply ascending integer values, not dates. The labels are then simply the texts from the dataframe.

An easy option to get rid of the hours and minutes is to reset the labels as follows:

ax = df.plot(kind='bar', stacked = False)
ax.set_xticklabels([t.get_text().split()[0] for t in ax.get_xticklabels()])

Additionally adding ax.figure.autofmt_xdate() rotates and recenters the labels to take less space.

edited Feb 25, 2017 at 12:20

answered Feb 25, 2017 at 12:15

ImportanceOfBeingErnest

342k61 gold badges737 silver badges771 bronze badges

Collectives™ on Stack Overflow

Python : Different behaviour of DatetimeIndex while plotting line and bar plots using DataFrame

2 Answers 2

1 Comment

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

1 Comment

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related