新的 Pandas 数据框列 - 每个 ID 和产品的最新日期

Question

I have a dataframe like this:

Index   ID       Item       Date
0       001      A          01/01/19
1       001      B          01/03/19
2       002      A          01/04/19
3       001      A          01/05/19
4       003      B          01/03/19
5       002      A          01/01/19

I would like to create a column that contains the latest date for every ID and Product. Currently, I am only able to get the latest date of all the dataset or the same date for every row with this code:

df['New Column Date'] = df['Date'].values[-1]

But the output should be like this:

Index   ID       Item      Date      New_column_date
0       001      A         01/01/19  NaN
1       001      B         01/03/19  NaN
2       002      A         01/04/19  NaN
3       001      A         01/05/19  01/01/19
4       003      B         01/03/19  NaN
5       002      A         01/01/19  01/04/2019  

Note: when we don´t have an earlier date, zero or NaN value.

Any solutions?

Answer 1

IIUC, you want groupby.shift :

df['new column']=df.groupby(['ID','Item'])['Date'].shift()
print(df)

   ID Item      Date new column
0   1    A  01/01/19        NaN
1   1    B  01/03/19        NaN
2   2    A  01/04/19        NaN
3   1    A  01/05/19   01/01/19
4   3    B  01/03/19        NaN
5   2    A  01/01/19   01/04/19

Answer 2

IIUC, we can use groupby + transform + max on your date column to get the latest date,

then filter by duplicates and apply the logic

only diff is that ID 1 for Item A shuld be 5th of Jan 2019?

s = df.groupby(['ID','Item'])['Date'].transform('max')
df.loc[df.duplicated(subset=['ID','Item']),'new_date'] = s
print(df)


   ID    Item       Date   new_date
0  001    A 2019-01-01        NaT
1  001    B 2019-01-03        NaT
2  002    A 2019-01-04        NaT
3  001    A 2019-01-05 2019-01-05
4  003    B 2019-01-03        NaT
5  002    A 2019-01-01 2019-01-04