how can I include the header?

Question

Quite new into pandas, I have generated a new column from division of last_value /first_value row in a this df.

df=
     data1 data2 
A      4    3
B      4    3
C      5    3
D      1    9

I generated this new column (with no headers)

new_column= df.loc[D]/df.loc[A]

new_column=
data1   0.25
data2   3

I have tried to add a header 'results' through :

new_column.rename(columns={0 :'results'}, inplace=True)
new_column.rename(columns={'0' :'results'}, inplace=True)
new_column['results']= df.loc[D]/df.loc[A]


print (new_column)

From the research done, nothing seems to work/adapt:

Adding Column Headers to new pandas dataframe

The result is the same without any headers.How can I include the header?

What I am looking for is :

       results
data1   0.25
data2   3

In short the new_column with the header 'results'

can you post your desired data set? it's not very clear what are you after... — MaxU - stand with Ukraine, Apr 21 '17 at 18:01

juanpa.arrivillaga · Answer 1 · 2017-04-21T18:19:35.547

new_column is a pd.Series, it doesn't have "columns", it does however have a name:

>>> new_column = df.loc['D']/df.loc['A']
>>> new_column
data1    0.25
data2    3.00
dtype: float64

By default, it doesn't have one if constructed this way, but:

>>> new_column.rename('results', inplace=True)
data1    0.25
data2    3.00
Name: results, dtype: float64

You want a pandas.DataFrame, so you could this instead of the above:

>>> new_column = (df.loc['D']/df.loc['A']).to_frame()
>>> new_column
          0
data1  0.25
data2  3.00

And now your attempts work:

>>> new_column.rename(columns={0 :'results'}, inplace=True)
>>> new_column
       results
data1     0.25
data2     3.00

Or, you could have done it in one, fell, swoop:

>>> new_column = pd.DataFrame((df.loc['D']/df.loc['A']), columns=['results'])
>>> new_column
       results
data1     0.25
data2     3.00

Edit to add suggestion by @MaxU

Or perhaps the easiest way is as suggested by MaxU:

>>> (df.loc['D']/df.loc['A']).to_frame('results')
       results
data1     0.25
data2     3.00

what about `(df.loc['D']/df.loc['A']).to_frame('results')`? ;-) — MaxU - stand with Ukraine, Apr 21 '17 at 18:17
@MaxU Nice, I was unaware `to_frame` took a `name` parameter! — juanpa.arrivillaga, Apr 21 '17 at 18:18

score 3 · Accepted Answer · edited Apr 21 '17 at 18:15

3

I'm editing this on my phone, so excuse the funky formatting, but does this do it?

new = pd.DataFrame({'results':df.loc[D]/df.loc[A]})

edited Apr 21 '17 at 18:15

mechanical_meat

163,903
24
228
223

answered Apr 21 '17 at 18:07

elPastor

8,435
11
53
81

great that is what I was looking for – JamesHudson81 Apr 21 '17 at 18:15

how can I include the header?

2 Answers2

Edit to add suggestion by @MaxU