如何在 pandas dataframe 中创建新列，该计算发生在除 go 之外的每一行之外

Question

For example let's say I got dataframe df with series A.1 and A.2 like so:

A.1    A.2
2      8
3      2
5      1

And I want to calculate let's say the difference of the means of all other rows like so:

A.1    A.2    B
2      8      (3+5)/2 - (2+1)/2
3      2      (2+5)/2-(8+1)/2
5      1      (2+3)/2-(8+2)/2

My code looks like this and doesn't work, how should I write it correctly?

df['B'] = mean(df['A.1'].drop(df['B'].index)))-mean(df['A.2'].drop(df['B'].index)))

I MUST totally avoid loops and do it in a panda-ish way as I'm working with huge datasets.

Answer 1

Try:

df.apply(lambda r : df.loc[df.index!=r.name,'A.1'].mean() - df.loc[df.index!=r.name,'A.2'].mean(), axis = 1)

result set is:

0    2.5
1   -1.0
2   -2.5
dtype: float64

Note that r.name inside lambda function is just index of current row.

Another approach with no lambda at all:

(df['A.1'].sum()-df['A.1'])/(len(df)-1) - (df['A.2'].sum()-df['A.2'])/(len(df)-1)

result is the same as above.