<p>我不知道你想要什么,但我会试一试。在</p>
<p>由于您没有提供数据,所以让我们创建四个具有不同漂移的随机行走:</p>
<pre><code>s1 = pd.Series(0.3 + np.random.normal(size=[100])).cumsum()
s2 = pd.Series(-0.3 + np.random.normal(size=[100])).cumsum()
s3 = pd.Series(0.1 + np.random.normal(size=[100])).cumsum()
s4 = pd.Series(0.1 + np.random.normal(size=[100])).cumsum()
</code></pre>
<p>以及<code>df</code>:</p>
^{pr2}$
<p>这样的情节</p>
<p><a href="https://i.stack.imgur.com/5HZpp.png" rel="nofollow noreferrer"><img src="https://i.stack.imgur.com/5HZpp.png" alt="enter image description here"/></a></p>
<p>现在,为了匹配最佳行,可以使用<a href="https://docs.scipy.org/doc/numpy-1.15.0/reference/generated/numpy.polyfit.html" rel="nofollow noreferrer">^{<cd2>}</a>将degree指定为1</p>
<pre><code>b1, a1 = np.polyfit(range(100), s1, 1)
b2, a2 = np.polyfit(range(100), s2, 1)
b3, a3 = np.polyfit(range(100), s3, 1)
b4, a4 = np.polyfit(range(100), s4, 1)
fig, ax = plt.subplots()
ax.plot(np.arange(100), a1 + b1*np.arange(100), color='red')
ax.plot(np.arange(100), a2 + b2*np.arange(100), color='blue')
ax.plot(np.arange(100), a3 + b3*np.arange(100), color='green')
ax.plot(np.arange(100), a4 + b4*np.arange(100), color='black')
</code></pre>
<p>以至于你</p>
<p><a href="https://i.stack.imgur.com/DLyhr.png" rel="nofollow noreferrer"><img src="https://i.stack.imgur.com/DLyhr.png" alt="enter image description here"/></a></p>
<p>要将最佳拟合线与实际初始图形进行比较,请在打印时设置相同的颜色:</p>
<pre><code>ax.plot(np.arange(100), a1 + b1*np.arange(100), color='red')
ax.plot(np.arange(100), a2 + b2*np.arange(100), color='blue')
ax.plot(np.arange(100), a3 + b3*np.arange(100), color='green')
ax.plot(np.arange(100), a4 + b4*np.arange(100), color='black')
ax.plot(df.s1, color='red')
ax.plot(df.s2, color='blue')
ax.plot(df.s3, color='green')
ax.plot(df.s4, color='black')
</code></pre>
<p><a href="https://i.stack.imgur.com/h5lK1.png" rel="nofollow noreferrer"><img src="https://i.stack.imgur.com/h5lK1.png" alt="enter image description here"/></a></p>