Let’s say that we have a data frame containing all purchases done by all our customers. We want to get the most recent purchase of every client. How do we do it?

We may use the grouping function to get the purchases of every customer and then get the most recent one from every group.

Imagine that the “group_name” column contains the identifier of the customer. The “used_for_sorting” column is our date of purchase, so we want the largest value in the group.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
import pandas as pd
data = pd.DataFrame([
    ['a', 1, 'A'],
    ['a', 2, 'B'],
    ['a', 3, 'C'],
    ['b', 5, 'D'],
    ['b', 6, 'E'],
    ['b', 7, 'F'],
    ['b', 8, 'G'],
    ['c', 10, 'H'],
    ['c', 11, 'I'],
    ['c', 12, 'J'],
    ['c', 13, 'K']
], columns = ['group_name', 'used_for_sorting', 'the_value'])

Get the first value from a group

In Pandas such a solution looks like that.
First, I have to sort the data frame by the “used_for_sorting” column.
I must do it before I start grouping because sorting of a grouped data frame is not supported and the groupby function does not sort the value within the groups, but it preserves the order of rows.

1
sorted_data_frame = data.sort_values(['used_for_sorting'], ascending=False)

Now, I can group the data frame by the customer identifier. In my case, the “group_name” is the customer identifier.

1
grouped_data_frame = sorted_data_frame.groupby('group_name')

After that, I must get the first value from every group. There are two options. I can either use the first function or the nth function with parameter 1. Obviously, I am going to choose the first function.

1
grouped_data_frame.first()

If you don’t want to use the group name as the index, remember to drop it.

All in one line, it looks like this:

1
2
3
4
5
data \
  .sort_values(['used_for_sorting'], ascending = False) \
  .groupby('group_name') \
  .first() \
  .reset_index()

The result:

1
2
3
4
     group_name  used_for_sorting the_value
0          a                 3         C
1          b                 8         G
2          c                13         K

Return the rank

What if instead of returning one row I want to get all of the rows with their rank?

In this case, I have to:

• group the data frame by the group_name column

• get the series (column) from the grouped data frame

• calculate the rank of the series

• add the rank as a new column in the original data frame

1
2
3
4
5
ranks = data \
  .groupby('group_name')['used_for_sorting'] \
  .rank(ascending = True, method = 'first')
ranks.name = 'rank'
pd.concat([data, ranks], axis = 1)

The result:

1
2
3
4
5
6
7
8
9
10
11
12
    group_name  used_for_sorting the_value  rank
0           a                 1         A   1.0
1           a                 2         B   2.0
2           a                 3         C   3.0
3           b                 5         D   1.0
4           b                 6         E   2.0
5           b                 7         F   3.0
6           b                 8         G   4.0
7           c                10         H   1.0
8           c                11         I   2.0
9           c                12         J   3.0
10          c                13         K   4.0