Question

I have multilevel dataframe ‘df’ like this :

             col1 col2
first second
a        0    5    5
         1    5    5
         2    5    5
b        0    5    5
         1    5    5

And I want to apply a function func (exp: 'lambda x: x*10') to second, somewhat like :

df.groupby(level='first').second.apply(func)

and result will lokk like:

             col1 col2
first second
a        0    5    5
         10   5    5
         20   5    5
b        0    5    5
         10   5    5

The above command not work for second is not a column, so .second is not accepted by Pandas .

I don’t want to do that by df.reset_index() , blablabla…, then finally df.set_index(). I prefer to do it in one command, How to do ?

Answer 1

When creating the DataFrame, you could set the MultiIndex as follows:

df.set_index(['first', 'second'], drop=False)

This way, the index column is not dropped and still accessible for your apply.