Pandas: Set specific columns to 0

Question

I have a Pandas dataframe with a single row. I want to simultaneously add a list of columns, called new_cols, and set all of their values to 0.

Is there a straightforward way of doing this?

My attempt:

article_features[new_cols] = 0 but that doesn't work :(

Thanks!

Please include a sample value of `new_cols` and the exact error message. — DYZ, Jul 31 '17 at 20:16
@DYZ Sorry, I was wrong: `new_cols` has a bunch of unicode in it — anon_swe, Jul 31 '17 at 20:20
@bclayman, can you post an output of `print(new_cols)` in your question? — MaxU - stand with Ukraine, Jul 31 '17 at 20:20
Unicode or string, same difference. But is `new_cols` a list of strings or one strings? If it is one string, where do other column names come from? — DYZ, Jul 31 '17 at 20:22
printing `new_cols[:10]` gives us `[u'utpica', u'cityvia', u'lostncheeseland', u'programcontribute', u'vang', u'deflatedlike', u'nunnery', u'vivemos', u'withamachete', u'gah']` — anon_swe, Jul 31 '17 at 20:24

MaxU - stand with Ukraine · Answer 1 · 2017-07-31T21:21:21.747

We can use assign method.

Demo:

In [43]: df
Out[43]:
   a  b  c
0  1  2  3
1  4  5  6

In [44]: new_cols
Out[44]: ['c1', 'c2', 'c3']

Option 1:

In [46]: df = df.assign(**{c:0 for c in new_cols})

In [47]: df
Out[47]:
   a  b  c  c1  c2  c3
0  1  2  3   0   0   0
1  4  5  6   0   0   0

Option 2:

In [106]: df = df.join(pd.DataFrame(0, df.index, new_cols))

In [107]: df
Out[107]:
   a  b  c  c1  c2  c3
0  1  2  3   0   0   0
1  4  5  6   0   0   0

adding string values:

In [49]: df = df.assign(**{c:'0' for c in new_cols})

In [50]: df
Out[50]:
   a  b  c c1 c2 c3
0  1  2  3  0  0  0
1  4  5  6  0  0  0

In [51]: df.dtypes
Out[51]:
a      int64
b      int64
c      int64
c1    object
c2    object
c3    object
dtype: object

The OP claims `new_cols` is a string. – DYZ Jul 31 '17 at 20:17 — DYZ, Jul 31 '17 at 20:17

score 2 · Accepted Answer · answered Jul 31 '17 at 21:18

Using pd.DataFrame.reindex_axis and the fill_value=0 parameter.

df.reindex_axis(df.columns.union(new_cols), axis=1, fill_value=0)

   a  b  c  c1  c2  c3
0  1  2  3   0   0   0
1  4  5  6   0   0   0

Or for strings use fill_value='0'

df.reindex_axis(df.columns.union(new_cols), 1, fill_value='0')

   a  b  c c1 c2 c3
0  1  2  3  0  0  0
1  4  5  6  0  0  0

Setup
I borrowed objects from @MaxU

df = pd.DataFrame({'a': [1, 4], 'b': [2, 5], 'c': [3, 6]})
new_cols = ['c1', 'c2', 'c3']

score 0 · Answer 3 · answered Jul 31 '17 at 20:29

0

You can add columns in a loop:

for col in new_cols:
    df[col] = 0

answered Jul 31 '17 at 20:29

DYZ

55,249
10
64
93

Pandas: Set specific columns to 0

3 Answers3