我正试图从ndarray
整数'标志'开始:
array([[1, 3, 2],
[2, 0, 3],
[3, 2, 0],
[2, 0, 1]])
到ndarray
个字符串:
array([['Banana', 'Celery', 'Carrot'],
['Carrot', 'Apple', 'Celery'],
['Celery', 'Carrot', 'Apple'],
['Carrot', 'Apple', 'Banana']],
dtype='|S6')
使用字符串列表作为'flags'到'含义'的映射:
meanings = ['Apple', 'Banana', 'Carrot', 'Celery']
我想出了以下内容:
>>> import numpy as np
>>> meanings = ['Apple', 'Banana', 'Carrot', 'Celery']
>>> flags = np.array([[1,3,2],[2,0,3],[3,2,0],[2,0,1]])
>>> flags
array([[1, 3, 2],
[2, 0, 3],
[3, 2, 0],
[2, 0, 1]])
>>> mapped = np.array([meanings[f] for f in flags.flatten()]).reshape(flags.shape)
>>> mapped
array([['Banana', 'Celery', 'Carrot'],
['Carrot', 'Apple', 'Celery'],
['Celery', 'Carrot', 'Apple'],
['Carrot', 'Apple', 'Banana']],
dtype='|S6')
这很有效,但在处理大型flatten
时,我关注相关行的效率(list comp,reshape
,ndarrays
):
np.array([meanings[f] for f in flags.flatten()]).reshape(flags.shape)
是否有更好/更有效的方式来执行这样的映射?
答案 0 :(得分:3)
花式索引是 numpythonic 的做法:
mapped = meanings[flags]
或通常更快的等价物:
mapped = np.take(meanings, flags)
答案 1 :(得分:2)
我认为np.vectorize
是要走的路,它也非常清晰易懂。我还没有对以下内容进行过测试,但它应该有效。
vfunc = np.vectorize(lambda x : meanings[x])
mapped = vfunc(flags)