idxmax() doesn't work on SeriesGroupBy that contains NaN

Question

Here is my code

from pandas import DataFrame, Series
import pandas as pd
import numpy as np
income = DataFrame({\'name\': [\'Adam\', \'Bill\', \'Chris\', \'Dave\         


        

Answer 1

Since groupby preserves order of rows within each group, you sort income before groupby. Then, pick up the firsts using head:

grouped=income.sort('income', ascending=False).groupby([ageBin])
highestIncome = income.ix[grouped.head(1).index]
#highestIncome is no longer ordered by age. 
#If you want to recover this, sort it again.
highestIncome.sort('age', inplace=True)

By the way, beware that the reference manual does not mention that groupby will preserve the order. I think most clean solution would be fix pandas's idxmax to work. For me, it is a little bit strange why idxmax does not work while max works.

Answer 2

Just apply a lambda function on the groups like so:

grouped.apply(lambda x: x.max())

Answer 3

grouped['income'].agg(lambda x : x.idxmax())


Out[]:
age
(20, 30]     1
(30, 40]   NaN
(40, 50]     2
(50, 60]     4
Name: income, dtype: float64

and then you can do the following to get the data

income.ix[result.values].dropna()