索引偏移的熊猫滚动功能

Question

The code

s = pd.Series([0,1,2,3,4])
sr = s.rolling(3)
sr.apply(np.sum)

returns the series with indices [0,1,2,3,4] and values [NaN, NaN, 3, 6, 9]. Is there a quick hack, specifically using pandas rolling functions, so that it returns the rolling sum from the following 3 indices, ie so that the series values are [3, 6, 9, NaN, NaN]?

Answer 1

The only difference is a shift by -2:

w = 3
s.rolling(w).sum().shift(-w + 1)

0    3.0
1    6.0
2    9.0
3    NaN
4    NaN
dtype: float64

Answer 2

Adding iloc[::-1]

s = pd.Series([0,1,2,3,4])
sr = s.iloc[::-1].rolling(3)
sr.sum().iloc[::-1]

0    3.0
1    6.0
2    9.0
3    NaN
4    NaN
dtype: float64

Answer 3

You need numpy.roll()

s = pd.Series([0,1,2,3,4])
sr = s.rolling(3)
pd.Series(np.roll(sr.apply(np.sum),3))

Output:

0    3.0
1    6.0
2    9.0
3    NaN
4    NaN
dtype: float64