Python:线性回归斜率和 y 轴截距

Question

我正在开发一个可以使用 linregress native scipyy 函数计算斜率的程序，但我遇到了两个错误(取决于我尝试修复它的方式)。这两个列表应该是二维的，基本上是 x 和 y 值。

from __future__ import division
from scipy.stats import linregress
import matplotlib.pyplot as mplot
import numpy as np

xs=[[20.0, 80.0, 45.0, 42.0, 93.0, 98.0, 65.0, 43.0, 72.0, 36.0, 9.0, 60.0, 47.0, 84.0, 31.0, 46.0, 57.0, 76.0, 27.0, 85.0, 0.0, 39.0, 2.0, 56.0, 68.0, 6.0, 41.0, 28.0, 61.0, 12.0, 32.0, 1.0, 54.0, 77.0, 18.0, 86.0, 62.0, 23.0, 30.0, 69.0, 4.0, 71.0, 64.0, 92.0, 24.0, 79.0, 8.0, 35.0, 49.0, 53.0, 7.0, 59.0, 70.0, 37.0, 13.0, 15.0, 73.0, 89.0, 96.0, 83.0, 22.0, 95.0, 19.0, 67.0, 5.0, 88.0, 38.0, 50.0, 55.0, 52.0, 81.0, 58.0, 11.0, 51.0, 99.0, 78.0, 25.0, 33.0, 40.0, 75.0, 3.0, 91.0, 48.0, 90.0, 82.0, 26.0, 10.0, 16.0, 21.0, 66.0, 14.0, 87.0, 74.0, 97.0, 94.0, 44.0, 29.0, 17.0, 63.0, 34.0], [87.0, 17.0, 69.0, 72.0, 76.0, 62.0, 20.0, 77.0, 5.0, 49.0, 81.0, 3.0, 24.0, 36.0, 44.0, 91.0, 99.0, 35.0, 43.0, 50.0, 12.0, 54.0, 46.0, 30.0, 37.0, 45.0, 90.0, 85.0, 70.0, 83.0, 38.0, 22.0, 23.0, 0.0, 60.0, 47.0, 26.0, 1.0, 95.0, 73.0, 65.0, 94.0, 84.0, 8.0, 34.0, 56.0, 66.0, 13.0, 75.0, 52.0, 19.0, 55.0, 67.0, 39.0, 21.0, 80.0, 98.0, 33.0, 11.0, 68.0, 40.0, 32.0, 2.0, 79.0, 82.0, 93.0, 96.0, 88.0, 14.0, 92.0, 41.0, 89.0, 28.0, 29.0, 42.0, 6.0, 86.0, 74.0, 58.0, 16.0, 31.0, 64.0, 15.0, 53.0, 25.0, 59.0, 61.0, 78.0, 51.0, 7.0, 57.0, 9.0, 97.0, 63.0, 48.0, 71.0, 18.0, 10.0, 4.0, 27.0]]

ys=[[155.506, 50.592, 104.447, 111.318, 36.148, 36.87, 74.266, 106.413, 58.341, 122.563, 180.555, 85.202, 96.84, 50.726, 126.56, 100.686, 88.303, 54.797, 138.487, 44.946, 200.9, 116.524, 193.652, 82.8, 65.823, 184.436, 113.738, 133.458, 83.765, 167.408, 129.491, 200.469, 89.238, 51.799, 159.217, 49.382, 78.443, 146.051, 129.045, 63.805, 185.564, 65.614, 74.243, 43.408, 140.863, 53.446, 182.767, 127.373, 94.494, 91.079, 187.194, 81.254, 68.702, 121.368, 164.756, 169.696, 59.483, 45.978, 33.057, 47.12, 154.755, 33.872, 160.754, 70.256, 190.393, 38.398, 113.188, 100.493, 84.511, 88.635, 49.353, 81.821, 178.876, 95.307, 32.2, 54.715, 141.389, 132.337, 109.673, 57.611, 189.251, 39.283, 97.31, 41.173, 47.529, 140.03, 173.058, 160.288, 154.773, 67.903, 164.718, 42.032, 60.739, 28.656, 34.302, 107.022, 137.344, 160.195, 73.636, 123.797], [14.138, 100.87, 30.287, 28.675, 21.826, 42.445, 97.938, 29.574, 125.976, 59.404, 26.609, 125.743, 95.329, 75.467, 59.497, 15.342, 9.834, 77.402, 65.019, 54.468, 112.64, 45.466, 55.197, 79.992, 71.146, 55.39, 14.795, 15.971, 28.535, 25.862, 73.239, 92.455, 87.635, 137.6, 38.59, 53.718, 86.26, 130.567, 11.274, 33.867, 40.035, 11.07, 16.109, 114.732, 76.552, 45.85, 31.827, 110.877, 26.292, 55.738, 101.801, 48.601, 33.632, 66.647, 98.39, 23.904, 11.172, 78.215, 109.417, 31.653, 68.368, 79.593, 124.548, 21.513, 19.828, 13.48, 9.993, 22.043, 108.229, 16.904, 66.704, 12.262, 79.947, 85.012, 66.754, 124.114, 17.548, 25.872, 45.392, 101.775, 78.085, 36.358, 101.795, 52.045, 87.637, 42.784, 37.011, 26.036, 50.146, 119.666, 42.514, 113.313, 9.125, 42.394, 51.954, 26.898, 96.678, 112.108, 125.252, 86.296]]

slope, intercept, r_value, std_err = linregress(xs,ys)
print(slope)

我的错误是:

  in linregress
    ssxm, ssxym, ssyxm, ssym = np.cov(x, y, bias=1).flat

ValueError: too many values to unpack (expected 4)

我试过将我的代码改成这样:

slope, intercept, r_value, std_err = linregress(xs[:,0], ys[:,0])

但随后我的错误变成了 TypeError:

TypeError: list indices must be integers or slices, not tuple

有人有什么建议吗？也许我对 linregress 函数的使用有一些不理解的地方。我确定我的第一个错误与我的列表是二维的有关。对于第二个错误，我迷路了。

Answer 1

你有两个问题:

• 当解释为数组时，您的变量 xs 和 ys 是二维的，形状为 (2, 100)。当 linregress 被赋予 x 和 y 这两个参数时，它期望它们是一维数组。

• 正如您在 the docstring 的“返回”部分中看到的那样, linregress 返回五个 值，而不是四个。

您必须调用 linregress 两次，并处理五个返回值。例如，

In [144]: slope, intercept, rvalue, pvalue, stderr = linregress(xs[0], ys[0])

In [145]: slope, intercept, rvalue
Out[145]: (-1.7059670627062702, 187.5658196039604, -0.9912859597363385)

In [146]: slope, intercept, rvalue, pvalue, stderr = linregress(xs[1], ys[1])

In [147]: slope, intercept, rvalue
Out[147]: (-1.2455432103210327, 121.51968891089112, -0.9871123119133126)