How to create a bar plot of the number of unique values within each group

Question:

import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns

## the following is the data set

gm = pd.read_csv('https://raw.githubusercontent.com/gerberl/6G7V0026-2223/main/datasets/gapminder.tsv', sep='t')

the command that I have been using, which counts each country multiple times.

sns.countplot(x=gm.continent)
plt.show

I can get the plot graph by making a new data frame, however there must be a way to get the graph without making a new dataframe.

The bars would be for the total number of countries in each continent, where the x-axis will be continents.

Asked By: saad1s

||
Source

Answers:

  • The most direct way is to use pandas to get the number of unique countries for each continent, and then plot directly with pandas.DataFrame.plot.
    • pandas uses matplotlib as the default plotting backend, and seaborn is just an API for matplotlib.
  • This answer shows how to use pd.DataFrame.pivot_table to get the number of unique values for each group.
    • gm.groupby('continent')['country'].nunique() can also be used.
  • If the link to the Gapminder data no longer works, it can also be found here.
import pandas as pd

# load the dataset
gm = pd.read_csv('https://raw.githubusercontent.com/gerberl/6G7V0026-2223/main/datasets/gapminder.tsv', sep='t')

# create a pivot table with continent and the number of unique countires
pt = gm.pivot_table(index='continent', values='country', aggfunc='nunique')

# plot the bar cart
ax = pt.plot(kind='bar', rot=0, ylabel='Number of Countires', xlabel='Continent', legend=False)

enter image description here

pt DataFrame

           country
continent         
Africa          52
Americas        25
Asia            33
Europe          30
Oceania          2
Answered By: Trenton McKinney

When one is building visualizations it is frequently useful to create additional variables.

For this specific case, one can start by creating a dataframe with the number of countries in each continent, by using pandas.core.groupby.DataFrameGroupBy.nunique and pandas.DataFrame.reset_index as follows

df_new = gm.groupby('continent')['country'].nunique().reset_index()

[Out]:

  continent  country
0    Africa       52
1  Americas       25
2      Asia       33
3    Europe       30
4   Oceania        2

Then, from that newly created dataframe df_new, one can create the bar plot using seaborn.barplot. Depending on the requirements can have a variety of adjustments. As OP didn’t specify strong opinions, the following will do the work

ax = sns.barplot(x='continent', y='country', data=df_new)
ax.set_title('Number of countries in each continent')
ax.set_xlabel('Continent')
ax.set_ylabel('Number of countries')
plt.show()

enter image description here

Answered By: Gonçalo Peres