import pandas as pd

data = {
    'city' : ['Toronto', 'Montreal', 'Waterloo', 'Toronto', 'Waterloo', 'Toronto', 'Toronto'],
    'points' : [80, 70, 90, 85, 79, 82, 200]
}

data

{'city': ['Toronto',
  'Montreal',
  'Waterloo',
  'Toronto',
  'Waterloo',
  'Toronto',
  'Toronto'],
 'points': [80, 70, 90, 85, 79, 82, 200]}

type(data)

dict

df = pd.DataFrame(data)

df

df.groupby('city')['points'].apply(lambda x:x.rolling(window=1).mean())

0     80.0
1     70.0
2     90.0
3     85.0
4     79.0
5     82.0
6    200.0
Name: points, dtype: float64

df.groupby('city')['points'].apply(lambda x:x.rolling(window=2).mean())

0      NaN
1      NaN
2      NaN
3     82.5
4     84.5
5     83.5
6    141.0
Name: points, dtype: float64