使用 Matplotlib 在半对数刻度上拟合直线

我一直在努力在用 Matplotlib 和 Python 3 制作的半对数图上拟合直线。我见过很多双对数比例尺数字的例子，但我尝试过的解决方案都不起作用（使用numpy）。线最终总会在某个地方弯曲。

以下是我迄今为止所掌握的：

import os
import matplotlib
import matplotlib.pyplot as plt
import numpy as np

base_path = os.path.dirname(os.path.realpath(__file__))

fig = plt.figure()
ax = fig.add_subplot(111)

# Plot data.
location = os.path.join(base_path, "data.csv")
data = np.genfromtxt(location, delimiter=',', names=['year', 'bw'])
ax.plot(data['year'], data['bw'])

# Fit test.
x = data['year']
y = data['bw']
y_ln = np.log10(y)

n = data.shape[0]
A = np.array(([[x[j], 1] for j in range(n)]))
B = np.array(y_ln[0:n])
B = np.array(y[0:n])

X = np.linalg.lstsq(A, B)[0]
a = X[0]
b = X[1]

fit = a * x + b

p = np.polyfit(x, np.log(y), 1)
ax.semilogy(x, p[0] * x + p[1], 'g--')

ax.set_yscale('log')

相关的data.csv文件如下所示：

2016, 68.41987090116676
2017, 88.9788618486191
2018, 90.94850458504749
2019, 113.20946182004333
2020, 115.71547492850719

The figure I obtain looks as following, where the fitted line is crooked.

非常感谢反馈和建议。

如果将数据的对数拟合到一条线上，则在实际绘制拟合数据时需要反转此操作。 IE。如果你适合一条线np.log(y)，你需要绘制np.exp(fit_result).

# Fit test.
x = data['year']
y = data['bw']

p = np.polyfit(x, np.log(y), 1)
ax.semilogy(x, np.exp(p[0] * x + p[1]), 'g--')

完整示例：

import io
import matplotlib.pyplot as plt
import numpy as np

u = u"""2016, 68.41987090116676
2017, 88.9788618486191
2018, 90.94850458504749
2019, 113.20946182004333
2020, 115.71547492850719"""

data = np.genfromtxt(io.StringIO(u), delimiter=',', names=['year', 'bw'])

fig = plt.figure()
ax = fig.add_subplot(111)
ax.plot(data['year'], data['bw'])

# Fit test.
x = data['year']
y = data['bw']

p = np.polyfit(x, np.log(y), 1)
ax.semilogy(x, np.exp(p[0] * x + p[1]), 'g--')

ax.set_yscale('log')

plt.show()