Numpy 如何重新排列行？

默认的

recarray

迭代器似乎正在遍历列。是否有一个对应于pandas的迭代行视图的

iterrows

/

itertuples

？当您使用时，

dtype

是什么？重新排列实际上没有形状。数据类型是

[（'fieldname'，'O'）]*n_字段

。

记录

具有数据类型，但

重新排列

，结构化数组上的变体也具有形状。当数组是2d时，列和行是有意义的。一个小例子可能会有所帮助。

In [102]: np.dtype('O,O,O')                                                     
Out[102]: dtype([('f0', 'O'), ('f1', 'O'), ('f2', 'O')])
In [103]: dt = np.dtype('O,O,O')                                                
In [104]: np.recarray((4,), dt)                                                 
Out[104]: 
rec.array([(None, None, None), (None, None, None), (None, None, None),
           (None, None, None)],
          dtype=[('f0', 'O'), ('f1', 'O'), ('f2', 'O')])
In [105]: arr = np.recarray((4,), dt)                                           
In [106]: arr                                                                   
Out[106]: 
rec.array([(None, None, None), (None, None, None), (None, None, None),
           (None, None, None)],
          dtype=[('f0', 'O'), ('f1', 'O'), ('f2', 'O')])
In [107]: arr.f0                                                                
Out[107]: array([None, None, None, None], dtype=object)
In [108]: arr['f0']                                                             
Out[108]: array([None, None, None, None], dtype=object)
In [109]: for i in arr: print(repr(i))                                          
(None, None, None)
(None, None, None)
(None, None, None)
(None, None, None)
In [110]: arr = np.recarray((4,1), dt)                                          
In [111]: arr                                                                   
Out[111]: 
rec.array([[(None, None, None)],
           [(None, None, None)],
           [(None, None, None)],
           [(None, None, None)]],
          dtype=[('f0', 'O'), ('f1', 'O'), ('f2', 'O')])
In [112]: for i in arr: print(repr(i))                                          
rec.array([(None, None, None)],
          dtype=[('f0', 'O'), ('f1', 'O'), ('f2', 'O')])
rec.array([(None, None, None)],
          dtype=[('f0', 'O'), ('f1', 'O'), ('f2', 'O')])
rec.array([(None, None, None)],
          dtype=[('f0', 'O'), ('f1', 'O'), ('f2', 'O')])
rec.array([(None, None, None)],
          dtype=[('f0', 'O'), ('f1', 'O'), ('f2', 'O')])