python - Pandas: Can you access rolling window items -

can access pandas rolling window object.

rs = pd.series(range(10)) rs.rolling(window = 3)  #print's  rolling [window=3,center=false,axis=0] 

can groups?:

[0,1,2] [1,2,3] [2,3,4] 

i start off saying reaching in internal impl. if really wanted compute indexers same way pandas.

you need v0.19.0rc1 (just released), can conda install -c pandas pandas=0.19.0rc1

in [41]: rs = pd.series(range(10))  in [42]: rs out[42]:  0    0 1    1 2    2 3    3 4    4 5    5 6    6 7    7 8    8 9    9 dtype: int64  # reaches internal implementation # first 3 window, second minimum periods # need     in [43]: start, end, _, _, _, _ = pandas._window.get_window_indexer(rs.values,3,3,none,use_mock=false)  # starting index in [44]: start out[44]: array([0, 0, 0, 1, 2, 3, 4, 5, 6, 7])  # ending index in [45]: end out[45]: array([ 1,  2,  3,  4,  5,  6,  7,  8,  9, 10])  # windo size in [3]: end-start out[3]: array([1, 2, 3, 3, 3, 3, 3, 3, 3, 3])  # indexers in [47]: [np.arange(s, e) s, e in zip(start, end)] out[47]:  [array([0]),  array([0, 1]),  array([0, 1, 2]),  array([1, 2, 3]),  array([2, 3, 4]),  array([3, 4, 5]),  array([4, 5, 6]),  array([5, 6, 7]),  array([6, 7, 8]),  array([7, 8, 9])] 

so sort of trivial in fixed window case, becomes extremely useful in variable window scenario, e.g. in 0.19.0 can specify things 2s example aggregate by-time.

all of said, getting these indexers not particularly useful. want do results. point of aggregation functions, or .apply if want generically aggregate.