如何根据行中的特定值和熊猫中的另一列对行进行分组？

Question

I can't think of a great way to ask this in one sentence so I'll show what I want to do.

Let's say I have table with each row being an event fired by someone going through a book rental process. We have 2 events, basket(contains the books they want) and checkout(fired when checkout is successful and books are rented). I want to group by name but also by checkout events and each basket event before it. Here's an example group just grouped on the name="tim".

|-------------------------------------------------|
| time | name |  stage   |        payload         |
|-------------------------------------------------|
| 1000 | tim  |  basket  | harrypotter;twilight;  |
|-------------------------------------------------|
| 1001 | tim  |  basket  | harrypotter;           |
|-------------------------------------------------|
| 1002 | tim  | checkout | Order# 123456789       |
|-------------------------------------------------|
| 1003 | tim  |  basket  | pandasfordummies;      |
|-------------------------------------------------|
| 1004 | tim  | checkout | Order# 145246263       |
|-------------------------------------------------|

My question is how can I group so that each group has 1 checkout event in it, like this:

First Order

|-------------------------------------------------|
| time | name |  stage   |        payload         |
|-------------------------------------------------|
| 1000 | tim  |  basket  | harrypotter;twilight;  |
|-------------------------------------------------|
| 1001 | tim  |  basket  | harrypotter;           |
|-------------------------------------------------|
| 1002 | tim  | checkout | Order# 123456789       |
|-------------------------------------------------|

Second Order

|-------------------------------------------------|
| time | name |  stage   |        payload         |
|-------------------------------------------------|
| 1003 | tim  |  basket  | pandasfordummies;      |
|-------------------------------------------------|
| 1004 | tim  | checkout | Order# 145246263       |
|-------------------------------------------------|

Sorry if this is worded terribly.

Answer 1

Assuming your table is in a pandas dataframe and already sorted by time and name, you can use the following code:

import numpy as np
import pandas as pd

df = pd.DataFrame({'time': [1000, 1001, 1002, 1003, 1004],
               'name':['tim', 'tim', 'tim', 'tim', 'tim'],
               'stage':['basket', 'basket', 'checkout', 'basket', 'checkout'],
               'payload':['harrypotter;twilight;', 'harrypotter;', 'Order# 123456789', 'pandasfordummies;', 'Order# 145246263']})

orders = np.split(df, np.where(df.stage == 'checkout')[0] + 1)

This will create an array of the split dataframes in orders that you can access normally like orders[0] , orders[1] etc.