Renaming X-Axis Labels when using Matplotlib and Pandas

Question

I'm running my code on iPython Notebooks, on a Macbook Pro Yosemite 10.10.4

I have a CSV file that I am trying to read using Python, and looking to come up with charts. The problem I am facing is renaming the X-Axis labels.

Essentially, the chart is trying to plot a count of different types of Audit Violations, but has really long descriptions of the said violations. For example:

• Not approved by regional committee.......another 300 words - 17
• No contract with vendor.......another 300 words - 14
• Vendor Registration not on record.......another 300 words - 9

Instead of having these verbose reasons though, I would like to rename the X-Axis labels to just numbers or alphabets so that the graph reads somewhat like this:

• A - 17
• B - 14
• C - 9

This is the code I have used, and except for the label names, I am happy with the result.

%matplotlib inline
import pandas as pd
import matplotlib.pyplot as plt
import numpy as np 
import pylab as pl
pd.set_option('display.mpl_style', 'default') 
pd.set_option('display.width', 5000) 
pd.set_option('display.max_columns', 60)
plt.rcParams['figure.figsize'] = (15, 5)
fixed_data = pd.read_csv('audit-rep.csv',sep=',',encoding='latin1',index_col='Index', parse_dates=['Report Date'],dayfirst=False)

viol_counts = data['Audit Flag'].value_counts()
viol_counts[:10]
viol_counts[:10].plot(kind='bar')

I have tried to rename the x-axis labels using the code below.

 viol_counts.set_ylabel('No. of Violations')
 viol_counts.set_title('Audit Results')
 viol_counts.set_xticks(ind+width)
 viol_counts.set_xticklabels( ('A', 'B','C') )

This is the error I get when using the above code.

AttributeError: 'Series' object has no attribute 'set_ylabel'

I have come across a few other posts related to this issue, but not seen one that specifically addresses the renaming of individual labels. This isn't utterly important though, and I'm just trying to learn using python, and the actual work has been done in excel.

Solution 1

if you assign the plot Axes object to a variable name (here I've called it viol_plot), then you can perform action on that Axes object (you are currently trying to set the labels and ticks on the Series, not the plot):

viol_plot = viol_counts[:10].plot(kind='bar')

viol_plot.set_ylabel('No. of Violations')
viol_plot.set_title('Audit Results')
viol_plot.set_xticks(ind+width)
viol_plot.set_xticklabels( ('A', 'B','C') )

Solution 2

Methods such as set_ylabel, set_title are for plot object. One option would be using subplot:

figure, ax = pl.subplots(1,1)
#note the ax option.
viol_counts[:10].plot(kind='bar', ax=ax)
ax.set_ylabel('No. of Violations')
ax.set_title('Audit Results')
ax.set_xticks(ind+width)
ax.set_xticklabels( ('A', 'B','C') )

Or, you can just plot and use different functions. One downside is that I don't know how to set xticklabels in this case:

viol_counts[:10].plot(kind='bar')
#note names of functions are slightly different
pl.ylabel('No. of Violations')
pl.title('Audit Results')
pl.xticks(ind+width)

Comments

[tacaswell]

viol_plot is an Axes object not a 'plot'.

[tacaswell]

Thanks. Not a huge fan of that API decision on the part of pandas (plotting functions should return the new artists added) but it is what it is.