How to make a histogram for non-numeric variables in python

Question

Sample data

import pandas as pd
import matplotlib.pyplot as plt

dummy = {'id': [1,2,3,4,5], 
        'brand': ['MS', 'Apple', 'MS', 'Google', 'Apple'], 
        'quarter': ['2017Q2', '2017Q2', '2017Q2', '2016Q1', '2015Q1']}

dummyData = pd.DataFrame(dummy, columns = ['id', 'brand', 'quarter'])
dummyData


# id    brand   quarter
# 0 1   MS      2017Q2
# 1 2   Apple   2017Q2
# 2 3   MS      2017Q2
# 3 4   Google  2016Q1
# 4 5   Apple   2015Q1

Now I want to plat a histogram using matplotlib and pandas, here the description

• X Axis : Quarter
• Y Axis : Count of values
• Histogram Bin: Filled with brand like 2017Q2 have two color values for MS and Apple
• Legends : Brand Name

I have a R background and its pretty easy using ggplot, I want to do the same in Python but I am not finding any suitable code, I am getting below mentioned error

TypeError: Empty 'DataFrame': no numeric data to plot

Answer 1

Another Alternative

data_frame.attribute_name.value_counts().plot.bar()

Example

iris_data.sample(3)

iris_data.Species.value_counts().plot.bar()

Answer 2

IIUC, you can use groupby + count + unstack + plot -

plt.style.use('ggplot')

dummyData.groupby(['quarter', 'brand'])\
      .brand.count().unstack().plot.bar(legend=True)

plt.show()

For reference, this is what is plotted -

brand    Apple  Google   MS
quarter                    
2015Q1     1.0     NaN  NaN
2016Q1     NaN     1.0  NaN
2017Q2     1.0     NaN  2.0