I would like fill a specif column of a dataset with the first not None value. For example, given the df:
col_name
None
None
A
A
B
B
output should be:
col_name
A
A
A
A
B
B
Any help on this, would be very appreciated. Thanks in advance. Carlo
2 Answers
df.update(df.col_name.iloc[:df.col_name.notnull().argmax() + 1].bfill())
df
col_name
0 A
1 A
2 A
3 A
4 B
5 B
Same Idea, but different
i = df.col_name.notnull().argmax()
df.col_name.values[:i] = df.col_name.values[i]
If None values are actually strings as in 'None'
i = df.col_name.ne('None').argmax()
df.col_name.iloc[:i] = df.col_name.iloc[i]
17 Comments
Carlo Allocca
how can I add inplace=True to your suggestion? Many Thanks,
Carlo Allocca
I see. I am running it...just a sec
Carlo Allocca
The first suggestion does not work for me. I try the other one.
Carlo Allocca
I got the following error: IndexError: index 173 is out of bounds for axis 0 with size 109
piRSquared
Are you sure those are
None and not 'None'. Can you run df.col_name.apply(type).iloc[:3] and report back what you get. |
You can convert the value to a numpy array by the dataframe.values variable, and then loop through it, comparing to numpy.nan, setting the value when true
eg.
colName = df['col_name'].values
colName[numpy.isnan(colName)] = 'A'
df.col_name.bfill()?df.col_name.apply(lambda x: numpy.nan if x==None else x)then do bfill