We might want to explore more and answer the question: For the schools with more black students than any other race, what is the distribution of its second highest race percentage?

>>> college_black = college_ugds[highest_percentage_race == 'UGDS_BLACK']
>>> college_black = college_black.drop('UGDS_BLACK', axis='columns')
>>> college_black.idxmax(axis='columns').value_counts(normalize=True)
UGDS_WHITE    0.661228
UGDS_HISP     0.230326
UGDS_UNKN     0.071977
UGDS_NRA      0.018234
UGDS_ASIAN    0.009597
UGDS_2MOR     0.006718
UGDS_AIAN     0.000960
UGDS_NHPI     0.000960
dtype: float64

We needed to drop the UGDS_BLACK column before applying the same method from this recipe. Interestingly, it seems that these schools with higher black populations have a tendency to have higher Hispanic populations.