如何按列值切片Numpy数组

时间:2011-09-28 01:37:07

标签: python numpy

我有一个像这个numpy数组的数组

 dd =[[0.567 2 0.611]
      [0.469 1 0.479]
      [0.220 2 0.269]
      [0.480 1 0.508]
      [0.324 1 0.324]]

我需要2个单独的数组dd[:,1] ==1dd[:,1] ==2

这些数组就是我追求的目标:

 na =[[0.469 1 0.479]
      [0.480 1 0.508]
      [0.324 1 0.324]]

 na2 =[[0.567 2 0.611]
       [0.220 2 0.269]]

我试过np.where确实有效

1 个答案:

答案 0 :(得分:6)

您可以使用numpy花式索引:

[~/repo/py]
|32>dd[dd[:,1] == 1]
[32] 
array([[ 0.469,  1.   ,  0.479],
       [ 0.48 ,  1.   ,  0.508],
       [ 0.324,  1.   ,  0.324]])

[~/repo/py]
|33>dd[dd[:,1] == 2]
[33] 
array([[ 0.567,  2.   ,  0.611],
       [ 0.22 ,  2.   ,  0.269]])

或者你可以使用列表理解:

[~/repo/py]
|21>np.array([row for row in dd if row[1] == 1])
[21] 
array([[ 0.469,  1.   ,  0.479],
       [ 0.48 ,  1.   ,  0.508],
       [ 0.324,  1.   ,  0.324]])

[~/repo/py]
|22>np.array([row for row in dd if row[1] == 2])
[22] 
array([[ 0.567,  2.   ,  0.611],
       [ 0.22 ,  2.   ,  0.269]])

编辑

如何在ipython中计算这些东西:

[~/repo/py]
|36>timeit dd[dd[:,1] == 1]
100000 loops, best of 3: 6 us per loop

[~/repo/py]
|37>timeit np.array([row for row in dd if row[1] == 1])
100000 loops, best of 3: 11.5 us per loop