How to keep one single column as a dataframe

Question

I have dataframe with 20 columns and one index.

Its shape is something like (100, 20).

I want to slice the 3rd column from this dataframe, but want to keep the result as a dataframe of (100,1).

If I do a v = df['col3'], I get a Series (which I do not want)
If I do a v =df[df['col3']!=0] and then v.drop(label=[list of 19 columns], axis = 1) --- I get what I want [that is a df(100,1)] but I have to

(a) write an unnecessary != condition ( which I want to avoid) and

(b) I have to write a long list of 19 column names.

There should be a better and cleaner way of doing what I wan to achieve.

Ami Tavory · Answer 1 · 2016-09-18T11:58:23.887

3

If I do a v = df['col3'], I get a Series (which I do not want)

If you use df[cols], where cols is a list, you'll get a DataFrame (not a Series). This includes the case where it is a list consisting of a single item. So, you can use df[['col3']].

For example:

In [33]: df = pd.DataFrame({'a': [1, 2], 'b': [3, 4]})

This gives a Series:

In [35]: df['a']
Out[35]: 
0    1
1    2
Name: a, dtype: int64

This gives a DataFrame:

In [36]: df[['a']]
Out[36]: 
   a
0  1
1  2

Finally, note that you can always transform a Series to a DataFrame with reset_index. So here, you can also use:

In [44]: df['a'].reset_index()
Out[44]: 
   index  a
0      0  1
1      1  2

edited Sep 18 '16 at 11:58

answered Sep 18 '16 at 08:32

Ami Tavory

74,578
11
141
185

Thanks a lot. But it looks like the second on returns a pandas.core.series.Series. – Prana Sep 18 '16 at 11:56
@Prana Sure. Do the two remaining options (erased that one) work for you? – Ami Tavory Sep 18 '16 at 11:58

score 3 · Accepted Answer · answered Sep 18 '16 at 09:03

3

Another handy trick is to_frame()

df['col3'].to_frame()

answered Sep 18 '16 at 09:03

piRSquared

285,575
57
475
624

Nice. Clearer in the code than the hackish-looking `reset_index()`. – Ami Tavory Sep 18 '16 at 11:15

How to keep one single column as a dataframe

2 Answers2