主页 / 问题 / 1176215
Accepted
Ildar
Asked:2020-09-09 20:09:05 +0000 UTC2020-09-09 20:09:05 +0000 UTC

离散经验分布函数 (cdf) 的构造

  • 772

构造离散分布函数时如何去除垂直线？

我目前的日程安排是什么： 在此处输入图像描述

import matplotlib.pyplot as plt


def plot_empirical_cdf(sample):
    plt.hist(sample, histtype='step', cumulative=True, bins=len(sample))
    plt.show()


sample = [123.0, 119.5, 120.5, 118.1, 117.5, 118.2, 124.5, 115.9, 121.0, 126.0, 121.3, 121.0, 118.2, 120.2, 124.3, 118.5, 117.2, 125.5, 115.0, 119.5, 126.0, 120.0, 126.5, 121.0, 125.0, 123.0, 122.7, 120.5, 122.5, 122.0, 122.5, 122.5, 122.5, 117.2, 123.0, 128.5, 120.5, 118.9, 116.2, 125.0, 125.1, 121.8, 122.0, 120.5, 123.0, 115.0, 120.5, 118.0, 125.0, 115.0, 117.0, 125.9, 122.0, 120.5, 121.1, 123.5, 120.7, 125.8, 120.5, 119.0, 120.5, 120.2, 123.0, 120.0, 117.2, 122.6, 120.2, 120.5, 123.0, 121.5, 120.0, 119.5, 115.0, 122.9, 123.3, 121.0, 122.5, 119.0, 124.0, 120.5, 121.9, 124.2, 121.5, 118.2, 122.2, 121.5, 120.0, 123.0, 119.5, 117.8, 119.2, 116.0, 123.0]
plot_empirical_cdf(sample)
python

2 个回答

  1. 任何建立在有限数量对象上的经验分布函数都将具有垂直步长。如果你真的需要 - 好吧，应用抗锯齿，但为什么呢？

    顺便说一句，没有必要自己构建 CDF 函数，statsmodels.api 中的 distributions.ECDF(sample) 函数在这方面做得很好

    但是如果你仍然需要一个平滑的函数，那么你可以这样做：

    import seaborn as sns
sample = [123.0, 119.5, 120.5, 118.1, 117.5, 118.2, 124.5, 115.9, 121.0, 126.0, 121.3, 121.0, 118.2, 120.2, 124.3, 118.5, 117.2, 125.5, 115.0, 119.5, 126.0, 120.0, 126.5, 121.0, 125.0, 123.0, 122.7, 120.5, 122.5, 122.0, 122.5, 122.5, 122.5, 117.2, 123.0, 128.5, 120.5, 118.9, 116.2, 125.0, 125.1, 121.8, 122.0, 120.5, 123.0, 115.0, 120.5, 118.0, 125.0, 115.0, 117.0, 125.9, 122.0, 120.5, 121.1, 123.5, 120.7, 125.8, 120.5, 119.0, 120.5, 120.2, 123.0, 120.0, 117.2, 122.6, 120.2, 120.5, 123.0, 121.5, 120.0, 119.5, 115.0, 122.9, 123.3, 121.0, 122.5, 119.0, 124.0, 120.5, 121.9, 124.2, 121.5, 118.2, 122.2, 121.5, 120.0, 123.0, 119.5, 117.8, 119.2, 116.0, 123.0]

sns.kdeplot(sample, cumulative=True)

    结果 在此处输入图像描述

    • 2
  2. Best Answer

    我可以建议你这个解决方案：

    import matplotlib.pyplot as plt
import numpy as np


def plot_empirical_cdf(sample):
    hist, edges = np.histogram(sample, bins=len(sample))
    Y = hist.cumsum()
    for i in range(len(Y)):
        plt.plot([edges[i], edges[i+1]],[Y[i], Y[i]], c="blue")
    plt.show()


sample = [123.0, 119.5, 120.5, 118.1, 117.5, 118.2, 124.5, 115.9, 121.0, 126.0, 121.3,
          121.0, 118.2, 120.2, 124.3, 118.5, 117.2, 125.5, 115.0, 119.5, 126.0, 120.0,
          126.5, 121.0, 125.0, 123.0, 122.7, 120.5, 122.5, 122.0, 122.5, 122.5, 122.5,
          117.2, 123.0, 128.5, 120.5, 118.9, 116.2, 125.0, 125.1, 121.8, 122.0, 120.5,
          123.0, 115.0, 120.5, 118.0, 125.0, 115.0, 117.0, 125.9, 122.0, 120.5, 121.1,
          123.5, 120.7, 125.8, 120.5, 119.0, 120.5, 120.2, 123.0, 120.0, 117.2, 122.6,
          120.2, 120.5, 123.0, 121.5, 120.0, 119.5, 115.0, 122.9, 123.3, 121.0, 122.5,
          119.0, 124.0, 120.5, 121.9, 124.2, 121.5, 118.2, 122.2, 121.5, 120.0, 123.0,
          119.5, 117.8, 119.2, 116.0, 123.0]
plot_empirical_cdf(sample)

    得到这张照片：

    在此处输入图像描述

    • 1

相关问题