Is it possible to create a bar chart where Y is the count of distinct values of a column in a Pandas data frame?

Question

For example, if my table is something like:

df = pd.DataFrame({'user_id': ['abc', 'abc', 'def'],
                   'product': ['A1', 'A2', 'A3']})

How can I return a bar chart where the dimension X is product and the metric y is the count of the distinct values of user_id?

The simplest way is to use `seaborn.countplot`: `sns.countplot(data=df, x='product', hue='user_id', dodge=False)` — Trenton McKinney, Aug 24 '21 at 15:54

mozway · Accepted Answer · 2021-08-24T14:50:37.293

You can groupby "product", count the unique values of "user_id" and plot:

(df.groupby('product')
   ['user_id'].nunique()
   .plot.bar()
)

output:

I find it a better example the other way around:

(df.groupby('user_id')
   ['product'].nunique()
   .plot.bar()
)

You might also be interested in identifying the groups (again reversed example as more demonstrative):

(df.groupby('user_id')
   ['product'].value_counts()
   .unstack()
   .plot.bar(stacked=True)
)

1 Answers1