Question

我有以下数组＆＃34; class_id＆＃34;

array([[array(['2'], 
  dtype='<U1')],
   [array(['3'], 
  dtype='<U1')],
   [array(['2'], 
  dtype='<U1')],
   ..., 
   [array(['2'], 
  dtype='<U1')],
   [array(['2'], 
  dtype='<U1')],
   [array(['2'], 
  dtype='<U1')]], dtype=object)

当我使用该功能时：

class_id.astype（int）

我收到以下错误：

   ValueError: setting an array element with a sequence.
---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-124-3ad39bf87e7d> in <module>()
----> 1 class_id.astype(int)
ValueError: setting an array element with a sequence.

我想获得以下数组：

array([[1],
   [2],
   [1],
   ..., 
   [1],
   [1],
   [1]])

我应该使用哪种功能？

Answer 1

您有一系列数组列表，您需要删除一个图层/维度以获得所需的输出;你可以试试这个：

import numpy as np
np.array([s[0].astype(int) for s in class_id])

#array([[2],
#       [3],
#       [2],
#       [2],
#       [2],
#       [2]])

class_id = array([[array(['2'], 
  dtype='<U1')],
   [array(['3'], 
  dtype='<U1')],
   [array(['2'], 
  dtype='<U1')],
   [array(['2'], 
  dtype='<U1')],
   [array(['2'], 
  dtype='<U1')],
   [array(['2'], 
  dtype='<U1')]], dtype=object)

Answer 2

要重新创建数组，我必须使用：

array = np.array
In [460]: alist = [[array(['2'], 
     ...:   dtype='<U1')],
     ...:    [array(['3'], 
     ...:   dtype='<U1')],
     ...:    [array(['2'], 
     ...:   dtype='<U1')],
     ...:    [array(['2'], 
     ...:   dtype='<U1')],
     ...:    [array(['2'], 
     ...:   dtype='<U1')],
     ...:    [array(['2'], 
     ...:   dtype='<U1')]]
In [461]: arr = np.zeros((6,1), dtype=object)
In [462]: arr[:,0] = alist
In [463]: arr
Out[463]: 
array([[[array(['2'], 
      dtype='<U1')]],
       [[array(['3'], 
      dtype='<U1')]],
       [[array(['2'], 
      dtype='<U1')]],
       [[array(['2'], 
      dtype='<U1')]],
       [[array(['2'], 
      dtype='<U1')]],
       [[array(['2'], 
      dtype='<U1')]]], dtype=object)

简单的复制n粘贴剥离内部数组定义：

array([[['2']],    # (6,1,1) shape

       [['3']],
       ...
       [['2']]], dtype=object)

使用arr[:,0]删除一级[]，我可以使用连接加入数组，并通过一次调用U1轻松将int数组转换为astype。

In [464]: np.concatenate(arr[:,0])
Out[464]: 
array([['2'],
       ['3'],
       ['2'],
       ['2'],
       ['2'],
       ['2']], 
      dtype='<U1')
In [465]: np.concatenate(arr[:,0]).astype(int)
Out[465]: 
array([[2],
       [3],
       [2],
       [2],
       [2],
       [2]])

在Python中将U1数组转换为int

2 个答案: