How can we add a flag to a field in a dataframe based on ID and descending date?

Question

I have the dataframe below.

import pandas as pd
import numpy as np

# data stored in dictionary
details = {
    'address_id': [111,111,111,111,111,111,222,222,222,222,222,222,333,333,333,333,333,333,444,444,444,444,444,444,555,555,555,555,555,555,777,777,777],
    'my_company':['Comcast','Verizon','Other','Other','Comcast','Comcast','Spectrum','Spectrum','Spectrum','Spectrum','Spectrum','Spectrum','Verizon','Verizon','Verizon','Verizon','Verizon','Verizon','Spectrum','Spectrum','Spectrum','Spectrum','Verizon','Spectrum','Spectrum','Spectrum','Spectrum','Spectrum','Verizon','Other','Verizon','Comcast','Comcast'],
    'my_date':['2022-01-24','2022-02-21','2022-03-28','2022-04-25','2022-05-23','2022-06-27','2022-01-24','2022-02-21','2022-03-28','2022-04-25','2022-05-23','2022-06-27','2022-01-24','2022-02-21','2022-03-28','2022-04-25','2022-05-23','2022-06-27','2022-01-24','2022-02-21','2022-03-28','2022-04-25','2022-05-23','2022-06-27','2022-01-24','2022-02-21','2022-03-28','2022-04-25','2022-05-23','2022-06-27','2022-01-24','2022-02-21','2022-03-28']
}

df = pd.DataFrame(details)
df

I want to see if an address_id switched AWAY from Verizon at some point during various durations of time, but grouped by address_id. If an address_id has Verizon in an earlier month, and switch to something else at any point later in time, I want to label it as a mover=TRUE.

If the house never had Verizon, then mover = False
If the house did not start with Verizon but ended with Verizon, then mover = False
If the house had Verizon at some point, but did not end with Verizon, then mover = True

I want to end up with this.

How can I do that?

So, I tried this…I think it’s close…

df['mover'] = df.address_id.map(df.groupby(['address_id'])['my_date']=descending).last().my_company) == 'Verizon'
df

I’m seeing: SyntaxError: invalid syntax

Asked By: ASH

||

Source

Answer 1

def f(ser):
    if 'Verizon' not in ser.unique():
        return False
    if ser.iloc[-1] == 'Verizon':
        return False
    return True

df['mover'] = df.groupby('address_id')['my_company'].transform(f)

Result:

    address_id my_company     my_date  mover
0          111    Comcast  2022-01-24   True
1          111    Verizon  2022-02-21   True
2          111      Other  2022-03-28   True
3          111      Other  2022-04-25   True
4          111    Comcast  2022-05-23   True
5          111    Comcast  2022-06-27   True
6          222   Spectrum  2022-01-24  False
7          222   Spectrum  2022-02-21  False
8          222   Spectrum  2022-03-28  False
9          222   Spectrum  2022-04-25  False
10         222   Spectrum  2022-05-23  False
11         222   Spectrum  2022-06-27  False
12         333    Verizon  2022-01-24  False
13         333    Verizon  2022-02-21  False
14         333    Verizon  2022-03-28  False
15         333    Verizon  2022-04-25  False
16         333    Verizon  2022-05-23  False
17         333    Verizon  2022-06-27  False
18         444   Spectrum  2022-01-24   True
19         444   Spectrum  2022-02-21   True
20         444   Spectrum  2022-03-28   True
21         444   Spectrum  2022-04-25   True
22         444    Verizon  2022-05-23   True
23         444   Spectrum  2022-06-27   True
24         555   Spectrum  2022-01-24   True
25         555   Spectrum  2022-02-21   True
26         555   Spectrum  2022-03-28   True
27         555   Spectrum  2022-04-25   True
28         555    Verizon  2022-05-23   True
29         555      Other  2022-06-27   True
30         777    Verizon  2022-01-24   True
31         777    Comcast  2022-02-21   True
32         777    Comcast  2022-03-28   True
33         888   Spectrum  2022-01-24  False
34         888    Comcast  2022-02-21  False
35         888    Verizon  2022-03-28  False

Answered By: Vladimir Fokow

How can we add a flag to a field in a dataframe based on ID and descending date?

Question:

Answers: