I have the following table:
| A | B | C |
|---|---|---|
| 0 | 0 | 0 |
| 2022-02-15 00:00:00 | 0 | 2022-03-46 00:00:00 |
| 2022-04-23 00:00:00 | 2021-01-10 00:00:00 | 0 |
I would like to get a resulting column that returns the column name of the latest date per row and ignore all empty rows (this is part of a bigger dataset so I can't dropna). So as follows:
| A | B | C | Result |
|---|---|---|---|
| 0 | 0 | 0 | 0 |
| 2022-02-15 00:00:00 | 0 | 2022-03-46 00:00:00 | C |
| 2022-04-23 00:00:00 | 2021-01-10 00:00:00 | 0 | A |
I converted the columns to datetime and tried using:
df['Result'] = df.idxmax(axis=1)
But this only resulted in the following error
TypeError: reduction operation 'argmax' not allowed for this dtype
CodePudding user response：
You need to have datetime type.
If you don't want to alter the dataframe, you can do it on the fly:
df['Result'] = df.apply(pd_to_datetime).idxmax(axis=1).mask(df.eq(0).all(1), 0)