熊猫-合并两个未堆叠的数据帧

Question

I'm working on a data set where I unstack the data, and get the first and last rows.

>>> print df.iloc[[(0), (-1)]]

     Distance               
Node            0     1     2   
Time                            
0             27.0  54.0  97.0  
60            22.0  49.0  92.0  

then I use df1 = (df.loc[0] - df.iloc[(-1)]).unstack() to get the difference of the two rows

Node            0    1    2    
Distance       5.0  5.0  5.0

And finally, I want to concatenate/combine the two data sets ( df and df1 ).

 >>> print pd.concat([df, df1], axis=1)

               (Distance, 0)  (Distance, 1)  (Distance, 2) 
0                   27.0           54.0           NaN  
60                  22.0           49.0           92.0 
Distance            NaN             NaN            NaN

but I'm getting NaN .

Does anyone know what I'm doing wrong?

I want my output to look like this

     Distance               
Node            0     1     2   
Time                            
0             27.0  54.0  97.0  
60            22.0  49.0  92.0 
Difference     5.0  5.0   5.0

Answer 1

In my opinion, the use of unstack is not needed for what you are trying to achieve. Below are a couple of alternatives. To simplify the formatting I've omitted index and column names.

Setup

df = pd.DataFrame({0: [27.0, 22.0], 1: [54.0, 49.0], 2: [97.0, 92.0]},
                  index=[0, 60])

print(df)

       0     1     2
0   27.0  54.0  97.0
60  22.0  49.0  92.0

pd.DataFrame.loc

Add a new row by index via pd.DataFrame.loc with a row label.

df.loc['Distance'] = df.loc[0] - df.iloc[-1]

pd.Series.to_frame + concat

Convert your difference series to a dataframe, then transpose and concatenate.

diff = (df.loc[0] - df.iloc[-1]).to_frame().T
diff.index = ['Distance']

res = pd.concat([df, diff])

Result

The result for both methods:

print(res)

             0     1     2
0         27.0  54.0  97.0
60        22.0  49.0  92.0
Distance   5.0   5.0   5.0