python - 如何在包含空列表的列表上垂直平均？

Question

我有一个包含以下内容的列表:

    list1 = [(4.974874129422414, 0.4384932775564907, 0.1879318517703546, 5.820735609514166, 0, 0), 
    (0.15069597326856923, 0.2961961688603689, 0.21595885700786707, 5.848923022691187, 1, 0), 
    (0.15085612758502492, 0.28850876174946627, 0.18977362640233908, 5.826501216543082, 0, 0), 
    (0.15069597326856923, 0.2887489932217097, 0.2176404773200905, 5.834028536994648, 1, 0), 
    (0.15093620474325167, 0.3005203353595069, 0.18961347208652674, 5.849643723630468, 0, 0), 
    (0.15069597326856923, 0.3235825566813912, 0.21515808543054254, 5.849964035159586, 1, 0), 
    (0.15085612758502492, 0.3520099475391594, 0.18937324061280378, 5.814569613228549, 0, 0), 
    (0.15093620474325167, 0.3860427394179732, 0.2174803230046498, 5.858131979266134, 1, 0), 
    (0.1506158961103403, 0.42768286128894817, 0.18969354924443318, 5.807843071967709, 0, 0)]

我需要这些值在垂直轴上的平均值，这样输出看起来像:

    [(average_col1, average_col2, average_col3, average_col4, average_col5, average_col6)]

但是，np.mean(list1, axis=1) 命令返回:

    IndexError: tuple index out of range

因此我尝试使用以下方法创建一个 numpy 数组:

    a = np.array(list1)
    a = array([ (4.974874129422414, 0.4384932775564907, 0.1879318517703546, 5.820735609514166, 0, 0),
   (0.15069597326856923, 0.2961961688603689, 0.21595885700786707, 5.848923022691187, 1, 0),
   (0.15085612758502492, 0.28850876174946627, 0.18977362640233908, 5.826501216543082, 0, 0),
   (0.15069597326856923, 0.2887489932217097, 0.2176404773200905, 5.834028536994648, 1, 0),
   (0.15093620474325167, 0.3005203353595069, 0.18961347208652674, 5.849643723630468, 0, 0),
   (0.15069597326856923, 0.3235825566813912, 0.21515808543054254, 5.849964035159586, 1, 0),
   (0.15085612758502492, 0.3520099475391594, 0.18937324061280378, 5.814569613228549, 0, 0),
   (0.15093620474325167, 0.3860427394179732, 0.2174803230046498, 5.858131979266134, 1, 0),
   (0.1506158961103403, 0.42768286128894817, 0.18969354924443318, 5.807843071967709, 0, 0)], 
  dtype=[('col1', '<f8'), ('col2', '<f8'), ('col3', '<f8'), ('col4', '<f8'), ('col5', '<i4'), ('col6', '<i4')])

如果我使用与上面相同的平均命令，它会返回:

    IndexError: tuple index out of range

因此我不确定从这里开始做什么。

Answer 1

您在使用 numpy 时遇到的问题是示例中矩阵的声明。

给定:

list1 = [(4.974874129422414, 0.4384932775564907, 0.1879318517703546, 5.820735609514166, 0, 0), 
    (0.15069597326856923, 0.2961961688603689, 0.21595885700786707, 5.848923022691187, 1, 0), 
    (0.15085612758502492, 0.28850876174946627, 0.18977362640233908, 5.826501216543082, 0, 0), 
    (0.15069597326856923, 0.2887489932217097, 0.2176404773200905, 5.834028536994648, 1, 0), 
    (0.15093620474325167, 0.3005203353595069, 0.18961347208652674, 5.849643723630468, 0, 0), 
    (0.15069597326856923, 0.3235825566813912, 0.21515808543054254, 5.849964035159586, 1, 0), 
    (0.15085612758502492, 0.3520099475391594, 0.18937324061280378, 5.814569613228549, 0, 0), 
    (0.15093620474325167, 0.3860427394179732, 0.2174803230046498, 5.858131979266134, 1, 0), 
    (0.1506158961103403, 0.42768286128894817, 0.18969354924443318, 5.807843071967709, 0, 0)]

您可以很容易地使用它在 numpy 中按列获取平均值:

>>> np.mean(list1, axis=0)
[ 0.68679585  0.34464285  0.20140261  5.83448231  0.44444444  0.        ]

接下来你有一个有趣的声明:

a = np.array([ (4.974874129422414, 0.4384932775564907, 0.1879318517703546, 5.820735609514166, 0, 0),
    (0.15069597326856923, 0.2961961688603689, 0.21595885700786707, 5.848923022691187, 1, 0),
    (0.15085612758502492, 0.28850876174946627, 0.18977362640233908, 5.826501216543082, 0, 0),
    (0.15069597326856923, 0.2887489932217097, 0.2176404773200905, 5.834028536994648, 1, 0),
    (0.15093620474325167, 0.3005203353595069, 0.18961347208652674, 5.849643723630468, 0, 0),
    (0.15069597326856923, 0.3235825566813912, 0.21515808543054254, 5.849964035159586, 1, 0),
    (0.15085612758502492, 0.3520099475391594, 0.18937324061280378, 5.814569613228549, 0, 0),
    (0.15093620474325167, 0.3860427394179732, 0.2174803230046498, 5.858131979266134, 1, 0),
    (0.1506158961103403, 0.42768286128894817, 0.18969354924443318, 5.807843071967709, 0, 0)], 
  dtype=[('col1', '<f8'), ('col2', '<f8'), ('col3', '<f8'), ('col4', '<f8'), ('col5', '<i4'), ('col6', '<i4')])

这与 matrix=np.array(list1) 不同它所做的是声明一个 numpy structured array 并命名每一列并为该列提供一个数据类型

该数组的每一行元素都是一个元组:

 >>> a[0]
 ( 4.97487413,  0.43849328,  0.18793185,  5.82073561, 0, 0)

而且您不能以通常的方式访问列:

>>> a[:,0]
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
IndexError: too many indices for array

因为它实际上是一个一维数组:

>>> a.shape
(9,)

相反，您必须按名称访问列:

>>> a['col1']
array([ 4.97487413,  0.15069597,  0.15085613,  0.15069597,  0.1509362 ,
        0.15069597,  0.15085613,  0.1509362 ,  0.1506159 ])

或者，按列名取平均值:

>>> [np.mean(a[col]) for col in ['col{}'.format(i) for i in range(1,7)]]
[0.68679584555500162, 0.34464284907500159, 0.20140260920884526, 5.8344823121106151, 0.44444444444444442, 0.0]