如何在python中对namedtuple使用sum（）/ average（）？

Question

from collections import namedtuple
Point = namedtuple('Point', ['x', 'y'])
points = [Point(x=1.0, y=1.0), Point(x=2.0, y=2.0)] 

我想计算points列表的平均点，即结果接收Point(1.5, 1.5)：

point = average(points) # x = 1.5, y = 1.5

例如。我知道如果np.average(points, axis=0)是points.shape，那就是(N, 2)，但我宁愿保留一个命名的元组。

Answer 1

计算平均坐标：

import numpy as np
Point(np.average([p.x for p in points]),
      np.average([p.y for p in points]))
#Point(x=1.5, y=1.5)

或者，更好地，隐式地将点列表转换为numpy数组，获得平均值，并将结果转换回Point

Point(*np.average(points, axis=0))
#Point(x=1.5, y=1.5)

Answer 2

也许我错过了什么，但如果你想避免numpy那么你可以做

>>> Point(sum(p.x for p in points)/len(points),sum(p.y for p in points)/len(points))
Point(x=1.5, y=1.5)

虽然看起来有点迂回。

Answer 3

这是namedtuple的一个新子类，它有一个classmethod来从一系列点的平均值或总和构造一个新点。

from collections import namedtuple


class Point(namedtuple('Point', ['x', 'y'])):
    @classmethod
    def from_average(cls, points):
        from_sum = cls.from_sum(points)

        return cls(from_sum.x / len(points), from_sum.y / len(points))

    @classmethod
    def from_sum(cls, points):
        if not all(isinstance(p, cls) for p in points):
            raise ValueError('All items in sequence must be of type {}'.format(cls.__name__))

        x = sum(p.x for p in points)
        y = sum(p.y for p in points)

        return cls(x, y)


point_1 = Point(1.0, 1.0)
point_2 = Point(2.0, 2.0)

point_3 = Point.from_average([point_1, point_2])

point_3
# Point(x=1.5, y=1.5)

point_4 = Point.from_sum([point_1, point_2])

point_4
# Point(x=3.0, y=3.0)

Answer 4

这是一种简洁的方式，只使用Python内置插件：

In [1]: from collections import namedtuple
   ...: Point = namedtuple('Point', ['x', 'y'])
   ...: points = [Point(x=1.0, y=1.0), Point(x=2.0, y=2.0)]
   ...:

In [2]: import statistics

In [3]: Point(*map(statistics.mean, zip(*points)))
Out[3]: Point(x=1.5, y=1.5)

你为什么一开始使用numpy？这里没有多大意义。