How to make a 3d array out of a 2d array, along the rows?-CodePudding

I have a 2d array that looks like this:

>>> X
array([[10, 15],
       [20, 25],
       [30, 35],
       [40, 45],
       [50, 55],
       [60, 65],
       [70, 75],
       [80, 85],
       [90, 95]])

>>> X.shape
(9, 2)

My goal is to make a new 3d array with 3 steps per every row:

>>> Y
array([[[10, 15],
        [20, 25],
        [30, 35]],

       [[20, 25],
        [30, 35],
        [40, 45]],

       [[30, 35],
        [40, 45],
        [50, 55]],

       [[40, 45],
        [50, 55],
        [60, 65]],

       [[50, 55],
        [60, 65],
        [70, 75]],

       [[60, 65],
        [70, 75],
        [80, 85]],

       [[70, 75],
        [80, 85],
        [90, 95]]])

>>> Y.shape
(7, 3, 2)

I'd like to know if there are any simple way to do this, without using loops.

CodePudding user response：

A pure numpy solution would be to use numpy.lib.stride_tricks.sliding_window_view:

from numpy.lib.stride_tricks import sliding_window_view

new_a = sliding_window_view(a, window_shape=(3,2)).reshape(7, 3, 2)

Output:

>>> new_a
array([[[10, 15],
        [20, 25],
        [30, 35]],

       [[20, 25],
        [30, 35],
        [40, 45]],

       [[30, 35],
        [40, 45],
        [50, 55]],

       [[40, 45],
        [50, 55],
        [60, 65]],

       [[50, 55],
        [60, 65],
        [70, 75]],

       [[60, 65],
        [70, 75],
        [80, 85]],

       [[70, 75],
        [80, 85],
        [90, 95]]])

CodePudding user response：

It's very easy to do with Pandas, using DataFrame.rolling:

df = pd.DataFrame(a)
new_a = np.array([w.to_numpy() for w in df.rolling(3) if len(w) == 3])

Output:

>>> new_a
array([[[10, 15],
        [20, 25],
        [30, 35]],

       [[20, 25],
        [30, 35],
        [40, 45]],

       [[30, 35],
        [40, 45],
        [50, 55]],

       [[40, 45],
        [50, 55],
        [60, 65]],

       [[50, 55],
        [60, 65],
        [70, 75]],

       [[60, 65],
        [70, 75],
        [80, 85]],

       [[70, 75],
        [80, 85],
        [90, 95]]])