Is possible to pass a DataFrame to an apply function like this?
df2 = df1.apply(func,axis=1,args=df2)
def func(df1,df2):
# do stuff in df2 for each row of df1
return df2
The two DataFrames do not have the same length.
Asked
Active
Viewed 3,632 times
1
Brad Solomon
- 38,521
- 31
- 149
- 235
Lorenzo Bottaccioli
- 441
- 1
- 7
- 20
-
I assume `return df2` should be indented? – Alexander Aug 03 '17 at 18:51
-
4`args=(df2, )` And yes it is possible. But there is a 90% chance of a better way to do this than what you're doing. – cs95 Aug 03 '17 at 18:52
-
@cᴏʟᴅsᴘᴇᴇᴅ thanks for the hint. Suggestions for better ways? – Lorenzo Bottaccioli Aug 03 '17 at 19:36
-
Can't say unless it is known what `func` is doing. Do you want to open a new question? – cs95 Aug 03 '17 at 19:37
-
It would help if you gave us some sample data according to: https://stackoverflow.com/help/mcve – Anton vBR Aug 03 '17 at 19:38
1 Answers
5
From the df.apply docs:
DataFrame.apply(func, axis=0, broadcast=False, raw=False, reduce=None, args=(), **kwds)Applies function along input axis of DataFrame.
args:tuplePositional arguments to pass to function in addition to the array/series.
The right way is to pass your arguments in a tuple, like this:
df1.apply(func, axis=1, args=(df2, ))
Further improvements to your code would be possible if it were known what you are trying to achieve.
cs95
- 379,657
- 97
- 704
- 746
-
Thanks very much for your answer. That solved a problem I was having. I had a function I wanted to apply to df1 which initially had no arguments, then I changed it to having a single df2 argument. So I went from `df1.apply(func, axis=1)` to `df1.apply(func, axis=1, args=(df2))` but that resulted in a warning about redundant parentheses so I changed it to `df1.apply(func, axis=1, args=df2)` which resulted in the error "The truth value of a DataFrame is ambiguous." So I changed it as you suggested to `df1.apply(func, axis=1, args=(df2, ))` and it is working just as intended. – MarkBC Mar 05 '21 at 22:30
-