Pivot 基于 pandas 中的特定条件的特定行到列

Question

This is the dataframe that I working with where there can be multiple customers associated with a certain case ID on different months (case_ID, cust_val,date are PK in this table).

case_ID| cust_val | date | primary | action | change   |
    1  | xx       | 3/2  |   1     |        | increase |
    1  | xx       | 3/2  |         |    1   | decrease |
    1  | xx       | 3/1  |   1     |        | decrease |
    1  | xx       | 3/1  |         |   1    | decrease |
    1  | yy       | 3/2  |   1     |        | decrease |
    1  | yy       | 3/2  |         |   1    | increase |
    2  | yy       | 3/2  |         |   1    | increase |
    2  | yy       | 3/2  |     1   |        | increase |

I want the output table to look like this, where for each case_ID, cust_val,date, the change associated with the primary and action are in one row:

case_ID| cust_val | date | primary_change | action_change |
    1  | xx       | 3/2  |   increase     |   decrease    |
    1  | xx       | 3/1  |   decrease     |   decrease    | 
    1  | yy       | 3/2  |   decrease     |   increase    | 
    2  | yy       | 3/2  |   increase     |   increase    | 

I tried this but this is obviosuly wrong and I am not sure how to solve this:

df.pivot(index=['case_ID','cust_val','date'], columns=['primary', 'action'], values='change').reset_index()

Any help is appreciated. Thanks in advance.

Answer 1

You can filter the dataframe and merge:

a = df[df.primary == "1"]  # <-- change "1" to 1 if the values are integers
b = df[df.action == "1"]

x = (
    pd.merge(a, b, on=["case_ID", "cust_val", "date"])
    .rename(columns={"change_x": "primary_change", "change_y": "action_change"})
    .drop(columns=["primary_x", "action_x", "primary_y", "action_y"])
)
print(x)

Prints:

   case_ID cust_val date primary_change action_change
0        1       xx  3/2       increase      decrease
1        1       xx  3/1       decrease      decrease
2        1       yy  3/2       decrease      increase
3        2       yy  3/2       increase      increase