similar to pandas.Series.rank()

[ztsweet]

import pandas as pd
#
df = pd.DataFrame({
    'name1':['A', 'B','A', 'B', 'A','B'],
    'name2':['a','b','c','d','e','f'],
    'goal':[92,91,90,89,88,87]
})
def f(x):
    x['rank_num'] = x['goal'].rank(ascending=False)
    return x
df1 = df.groupby('name1').apply(lambda x:f(x)).sort_values(['name1', 'name2'])
print(df1)

`
  | name1 | name2 | goal | rank_num
-- | -- | -- | -- | --
A | a | 92 | 1.0
A | c | 90 | 2.0
A | e | 88 | 3.0
B | b | 91 | 1.0
B | d | 89 | 2.0
B | f | 87 | 3.0
`

[oleksiyskononenko]

@samukweku Can we move the rank() implementation to this issue and close #3279?