使用 numpy 拟合数据

问题描述 投票:0回答:3

我有以下数据:

>>> x
array([ 3.08,  3.1 ,  3.12,  3.14,  3.16,  3.18,  3.2 ,  3.22,  3.24,
    3.26,  3.28,  3.3 ,  3.32,  3.34,  3.36,  3.38,  3.4 ,  3.42,
    3.44,  3.46,  3.48,  3.5 ,  3.52,  3.54,  3.56,  3.58,  3.6 ,
    3.62,  3.64,  3.66,  3.68])

>>> y
array([ 0.000857,  0.001182,  0.001619,  0.002113,  0.002702,  0.003351,
    0.004062,  0.004754,  0.00546 ,  0.006183,  0.006816,  0.007362,
    0.007844,  0.008207,  0.008474,  0.008541,  0.008539,  0.008445,
    0.008251,  0.007974,  0.007608,  0.007193,  0.006752,  0.006269,
    0.005799,  0.005302,  0.004822,  0.004339,  0.00391 ,  0.003481,
    0.003095])

现在,我想用 4 次多项式来拟合这些数据。所以我这样做:

>>> coefs = np.polynomial.polynomial.polyfit(x, y, 4)
>>> ffit = np.poly1d(coefs)

现在我为 x 值创建一个新网格来评估拟合函数

ffit
:

>>> x_new = np.linspace(x[0], x[-1], num=len(x)*10)

当我使用命令完成所有绘图(数据集和拟合曲线)时:

>>> fig1 = plt.figure()                                                                                           
>>> ax1 = fig1.add_subplot(111)                                                                                   
>>> ax1.scatter(x, y, facecolors='None')                                                                     
>>> ax1.plot(x_new, ffit(x_new))                                                                     
>>> plt.show()

我得到以下信息:

fitting_data.png fitting_data.png

我期望的是拟合函数能够正确拟合(至少接近数据的最大值)。我做错了什么?

python numpy regression curve-fitting data-fitting
3个回答
115
投票

不幸的是,

np.polynomial.polynomial.polyfit
以与
np.polyfit
np.polyval
相反的顺序返回系数(或者,如您使用的
np.poly1d
)。举例说明:

In [40]: np.polynomial.polynomial.polyfit(x, y, 4)
Out[40]: 
array([  84.29340848, -100.53595376,   44.83281408,   -8.85931101,
          0.65459882])

In [41]: np.polyfit(x, y, 4)
Out[41]: 
array([   0.65459882,   -8.859311  ,   44.83281407, -100.53595375,
         84.29340846])

一般来说:

np.polynomial.polynomial.polyfit
返回系数
[A, B, C]
A + Bx + Cx^2 + ...
,而
np.polyfit
返回:
... + Ax^2 + Bx + C

所以如果你想使用这个函数组合,你必须颠倒系数的顺序,如下所示:

ffit = np.polyval(coefs[::-1], x_new)

但是,文档明确指出要避免

np.polyfit
np.polyval
np.poly1d
,而只使用新的(呃)包。

仅使用多项式包是最安全的:

import numpy.polynomial.polynomial as poly

coefs = poly.polyfit(x, y, 4)
ffit = poly.polyval(x_new, coefs)
plt.plot(x_new, ffit)

或者,创建多项式函数:

ffit = poly.Polynomial(coefs)    # instead of np.poly1d
plt.plot(x_new, ffit(x_new))

fit and data plot


37
投票

请注意,您可以直接使用 Polynomial 类进行拟合并返回 Polynomial 实例。

from numpy.polynomial import Polynomial

p = Polynomial.fit(x, y, 4)
plt.plot(*p.linspace())

p
使用缩放和移动的 x 值来实现数值稳定性。如果您需要系数的通常形式,则需要遵循

pnormal = p.convert(domain=(-1, 1))

2
投票

使用 numpy 和 matplotlib 使用切比雪夫级数和多项式级数最小二乘最佳拟合曲线拟合数据

快速总结

执行数据拟合时需要注意的关键行(在下面的完整代码中)是,例如:

import matplotlib.pyplot as plt
from numpy.polynomial import Polynomial

# ...

cheby_series = Chebyshev.fit(x, y, deg=5)
x_cheby, y_cheby = cheby_series.linspace()

poly_series = Polynomial.fit(x, y, deg=5)
x_poly, y_poly = poly_series.linspace()

# ...

plt.plot(x_cheby, y_cheby, linewidth=6, alpha=0.5,
   label="Chebyshev Series 5th degree\nleast squares best fit curve")
plt.plot(x_poly, y_poly, 'k', linewidth=1,
   label="Polynomial Series 5th degree\nleast squares best fit curve")

这是我的完整程序的输出如下不是来自上面的部分程序片段):

详情

即使在阅读了这里的所有答案之后,由于 Matplotlib 的变化以及我遇到的一些其他问题,我真的一直在努力解决这个问题,直到我终于找到了这个演示!:https://numpy.org/doc/stable /reference/routines.polynomials.classes.html.

它在最底部的“配件”部分下包含一个很好的示例!:

import numpy as np
import matplotlib.pyplot as plt
from numpy.polynomial import Chebyshev as T

np.random.seed(11)
x = np.linspace(0, 2*np.pi, 20)
y = np.sin(x) + np.random.normal(scale=.1, size=x.shape)
p = T.fit(x, y, 5)
plt.plot(x, y, 'o')
xx, yy = p.linspace()
plt.plot(xx, yy, lw=2)
p.domain
p.window
plt.show()

这是它的增强版本,包含我的修改和一些有用的评论。我从我的 eRCaGuy_hello_world 存储库中的 plot_best_fit_polynomial.py 演示修改了下面的代码:

import matplotlib.pyplot as plt
from numpy.polynomial import Chebyshev
from numpy.polynomial import Polynomial

# data to fit
x =   [0.        , 0.33069396, 0.66138793, 0.99208189, 1.32277585,
       1.65346982, 1.98416378, 2.31485774, 2.64555171, 2.97624567,
       3.30693964, 3.6376336 , 3.96832756, 4.29902153, 4.62971549,
       4.96040945, 5.29110342, 5.62179738, 5.95249134, 6.28318531]
y =   [ 0.17494547,  0.29609217,  0.5657562 ,  0.57183462,  0.9685718 ,
        0.96462136,  0.86211039,  0.76726418,  0.51805246,  0.05803429,
       -0.25321856, -0.52352074, -0.66675568, -0.85965411, -1.12713934,
       -1.08134779, -0.76348274, -0.45674931, -0.32780698, -0.06834466]

# Obtain a 5th degree (order) least-squares fit curve to the x, y data using a
# Chebyshev Series
cheby_series = Chebyshev.fit(x, y, deg=5)
# Lines-space: get evenly-spaced points to plot a line; see:
# https://numpy.org/doc/stable/reference/generated/numpy.polynomial.chebyshev.Chebyshev.linspace.html
x_cheby, y_cheby = cheby_series.linspace()

# Now do the same things with a 5th order Polynomial Series fit as well!
# see: https://numpy.org/doc/stable/reference/generated/numpy.polynomial.polynomial.Polynomial.html
poly_series = Polynomial.fit(x, y, deg=5)
x_poly, y_poly = poly_series.linspace()

# plot all the data

f1 = plt.figure()

plt.plot(x, y, 'o')
plt.plot(x_cheby, y_cheby, linewidth=6, alpha=0.5,
   label="Chebyshev Series 5th degree\nleast squares best fit curve")
plt.plot(x_poly, y_poly, 'k', linewidth=1,
   label="Polynomial Series 5th degree\nleast squares best fit curve")

plt.legend()
plt.show()

相关:

  1. 我刚刚弄清楚如何标记图:如何在 Matplotlib 中添加图标题、图副标题、图页脚、图标题、轴标签、图例标签和 (x, y) 点标签

参考资料:

  1. https://numpy.org/doc/stable/reference/routines.polynomials.classes.html - 请参阅最底部的演示
  2. https://numpy.org/doc/stable/reference/routines.polynomials.html#documentation-for-the-polynomial-package
  3. https://numpy.org/doc/stable/reference/routines.polynomials.chebyshev.html
  4. https://numpy.org/doc/stable/reference/ generated/numpy.polynomial.chebyshev.Chebyshev.linspace.html
  5. https://numpy.org/doc/stable/reference/ generated/numpy.polynomial.polynomial.Polynomial.html
© www.soinside.com 2019 - 2024. All rights reserved.