Question

我有一个向量列表：

>>> import numpy as np
>>> num_dim, num_data = 10, 5
>>> data = np.random.rand(num_data, num_dim)
>>> data
array([[ 0.0498063 ,  0.18659463,  0.30563225,  0.99681495,  0.35692358,
         0.47759707,  0.85755606,  0.39373145,  0.54677259,  0.5168117 ],
       [ 0.18034536,  0.25935541,  0.79718771,  0.28604057,  0.17165293,
         0.90277904,  0.94016733,  0.15689765,  0.79758063,  0.41250143],
       [ 0.80716045,  0.84998745,  0.17893211,  0.36206016,  0.69604008,
         0.27249491,  0.92570247,  0.446499  ,  0.34424945,  0.08576628],
       [ 0.35311449,  0.67901964,  0.71023927,  0.03120829,  0.72864953,
         0.60717032,  0.8020118 ,  0.36047207,  0.46362718,  0.12441942],
       [ 0.1955419 ,  0.02702753,  0.76828842,  0.5438226 ,  0.69407709,
         0.20865243,  0.12783666,  0.81486189,  0.95583274,  0.30157658]])

从data，我需要随机选择3个向量，我可以用：

>>> import random
>>> random.sample(data, 3)
[array([ 0.80716045,  0.84998745,  0.17893211,  0.36206016,  0.69604008,
        0.27249491,  0.92570247,  0.446499  ,  0.34424945,  0.08576628]), array([ 0.18034536,  0.25935541,  0.79718771,  0.28604057,  0.17165293,
        0.90277904,  0.94016733,  0.15689765,  0.79758063,  0.41250143]), array([ 0.35311449,  0.67901964,  0.71023927,  0.03120829,  0.72864953,
        0.60717032,  0.8020118 ,  0.36047207,  0.46362718,  0.12441942])]

我已在http://docs.scipy.org/doc/numpy/reference/routines.random.html检查了文档，但我无法确定numpy中是否存在random.sample()这样的功能。

numpy.random.sample()是否与random.sample()相同？

random.sample()中的numpy是否相等？

Answer 1

正如@ayhan所证实的那样，可以这样做：

>>> data[np.random.choice(len(data), size=3, replace=False)]
array([[ 0.80716045,  0.84998745,  0.17893211,  0.36206016,  0.69604008,
         0.27249491,  0.92570247,  0.446499  ,  0.34424945,  0.08576628],
       [ 0.35311449,  0.67901964,  0.71023927,  0.03120829,  0.72864953,
         0.60717032,  0.8020118 ,  0.36047207,  0.46362718,  0.12441942],
       [ 0.1955419 ,  0.02702753,  0.76828842,  0.5438226 ,  0.69407709,
         0.20865243,  0.12783666,  0.81486189,  0.95583274,  0.30157658]])

来自docs：

numpy.random.choice（a，size = None，replace = True，p = None）

从给定的1-D数组生成随机样本

np.random.choice(data, size=3, replace=False)从data的索引列表中选择3个元素，而无需替换。

然后data[...]对索引进行切片并检索使用np.random.choice选择的索引。

如何使用numpy从列表中随机选择n个元素？

1 个答案: