How can we find new IDs added, or dropped, each month?-CodePudding

How can we find new IDs added per month? I have this dataframe.

import pandas as pd
import numpy as np
 
# data stored in dictionary
details = {
    'address_id': [111, 111, 111, 111, 222, 222, 222, 333, 333, 444, 444, 555, 555, 777],
    'mydate':['2022-01-24', '2022-01-24', '2022-03-28', '2022-03-28', '2022-01-24', '2022-01-24', '2022-03-28', '2022-01-24', '2022-03-28', '2022-01-24', '2022-03-28', '2022-01-24', '2022-04-16', '2022-03-28']
}
 
df = pd.DataFrame(details)
df

I can easily group by date and find unique IDs

df_id = df.groupby('mydate').address_id.nunique().reset_index()
df_id

But how can I list out newly added IDs and recently dropped IDs? How can I create a list, or dataframe, showing me that 555 did not exist for '2022-03-28', so this is an add. Also, 111, 222, 333, 444, and 777 were all dropped for '2022-04-16'

CodePudding user response：

Create a set for each day:

daily_ids = df.groupby('mydate')['address_id'].agg(set).sort_index(ascending=False)
print(daily_ids)

# Output:

mydate
2022-04-16                        {555}
2022-03-28    {777, 333, 111, 444, 222}
2022-01-24    {555, 333, 111, 444, 222}
Name: address_id, dtype: object

Now we can check the differences by looking at the differences between sets.

# Recently Removed:
>>> daily_ids.diff().shift(-1)
mydate
2022-04-16    {777, 333, 111, 444, 222}
2022-03-28                        {555}
2022-01-24                          NaN
Name: address_id, dtype: object

# Recently Added:
>>> daily_ids.diff(-1)
mydate
2022-04-16    {555}
2022-03-28    {777}
2022-01-24      NaN
Name: address_id, dtype: object

CodePudding user response：

df_id['unique'] = df.groupby('mydate')['address_id'].agg(set).set_axis(df_id.index)
df_id['new'] = sets_of_ids['address_id'].diff()
df_id['dropped'] = sets_of_ids['address_id'].diff(-1).shift()

df_id

Result:

       mydate  address_id                     unique    new                   dropped  
0  2022-01-24           5  {555, 333, 111, 444, 222}    NaN                       NaN 
1  2022-03-28           5  {777, 333, 111, 444, 222}  {777}                      {555} 
2  2022-04-16           1                      {555}  {555}  {777, 333, 111, 444, 222}