对绘图中的拟合设置限制

Question

我有一个数据图，并对其应用了线性拟合，但是我不确定为什么，但拟合线远离数据本身。我将如何对这条线施加限制，使其恰好适合我的数据（最好也让数据成为图表的焦点！）

图形输出和代码如下：

plt.plot(np.log(ts), amps, "1", ms=5, label="Schmitt Analysis (Amplitude against Log(Time))")

##Plot Linear Fit
y1, r, *_ = np.polyfit(amps, np.log(ts), 1, full=True)
f1 = np.poly1d(y1)
plt.plot(amps, f1(amps), label=f"linear ($\chi^2$ = {r[0]:0.2f})")

plt.xlabel("Log(Time)")
plt.ylabel("Amplitude")
plt.title("Schmitt Analysis (Amplitude against Log(Time))")
plt.xlim(0,10)
plt.ylim(-40,80)
plt.legend()

plt.savefig('A_Schmitt.jpg')

实际使用的数据：

log(ts) = [-inf 2.89037176 3.58351894 3.98898405 4.49980967 4.68213123 4.83628191 4.9698133 5.08759634 5.19295685 5.28826703 5.37527841 5.45532112 5.52942909 5.59842196 5.7235851 5.78074352 5.83481074 5.9348942 6.02586597 6.06842559 6.10924758 6.1484683 6.22257627 6.25766759 6.32435896 6.38687932 6.41673228 6.44571982 6.50128967 6.52795792 6.5539334 6.71901315 6.78219206]

安培= [77.78630383833547，62.92926582239441，63.84025706577048，55.489066870438165，38.60797989548756，40.771390484048545，14.679073842876978，33.95959972488966，29.41960790300141，32.93241034391399，30.927428194781815，31.086396885182356，21.52771899125612，4.27684299160886，6.432975528727562，7.500376934048583，18.730555740591637，4.355896959987761，11.677509915219987，12.865482314301719，0.6120306267606219，12.614420497451556， 2.2025029753442404，9.447046999592711，4.0688197216393425，0.546672901996845，1.12780050608251，2.2030852358874635，2.202804718915858，0.5726686031033587，0.5465322281618783，0.5185100682386156，0.575055917739342，0.5681697592593679]

注意到我犯了一个错误，我设法让图表更新了一点，但现在拟合完全失败了。

我也将上面的代码更新到了新版本。

Answer 1

这里有两个问题——一个错字和-inf的存在。
首先是拼写错误 - 您将 logts 和 amps 分别归为 y 和 x，符合您的要求，而应该反过来。
其次，对数转换的时间数组中 -inf 的存在没有被 fit 例程很好地处理。我们可以用logts[1:]手动排除第一个点。

import numpy as np
from matplotlib import pyplot as plt

#recreating your input - seemingly log(ts) is a numpy array
logts = np.asarray([-np.inf, 2.89037176, 3.58351894, 3.98898405, 4.49980967, 4.68213123, 4.83628191, 4.9698133, 5.08759634, 5.19295685, 5.28826703, 5.37527841, 5.45532112, 5.52942909, 5.59842196, 5.7235851, 5.78074352, 5.83481074, 5.9348942, 6.02586597, 6.06842559, 6.10924758, 6.1484683, 6.22257627, 6.25766759, 6.32435896, 6.38687932, 6.41673228, 6.44571982, 6.50128967, 6.52795792, 6.5539334, 6.71901315, 6.78219206])
amps = [77.78630383833547, 62.92926582239441, 63.84025706577048, 55.489066870438165, 38.60797989548756, 40.771390484048545, 14.679073842876978, 33.95959972488966, 29.41960790300141, 32.93241034391399, 30.927428194781815, 31.086396885182356, 21.52771899125612, 4.27684299160886, 6.432975528727562, 7.500376934048583, 18.730555740591637, 4.355896959987761, 11.677509915219987, 12.865482314301719, 0.6120306267606219, 12.614420497451556, 2.2025029753442404, 9.447046999592711, 4.0688197216393425, 0.546672901996845, 1.12780050608251, 2.2030852358874635, 2.202804718915858, 0.5726686031033587, 0.5465322281618783, 0.5185100682386156, 0.575055917739342, 0.5681697592593679]

#plot raw data
plt.plot(logts, amps, "1", ms=5, label="Schmitt Analysis (Amplitude against Log(Time))")

#linear fit excluding the first point that is an outlier
y1, r, *_ = np.polyfit(logts[1:], amps[1:], 1, full=True)
f1 = np.poly1d(y1)

#get min and max of logts excluding nan and inf values
logtsmin = np.floor(np.nanmin(logts[logts != -np.inf]))
logtsmax = np.ceil(np.nanmax(logts[logts != np.inf]))
#evenly spaced x-values for the fit line plot 
xlogts = np.linspace(logtsmin, logtsmax, 1000)
plt.plot(xlogts, f1(xlogts), label=f"linear ($\chi^2$ = {r[0]:0.2f})")

plt.xlabel("Log(Time)")
plt.ylabel("Amplitude")
plt.title("Schmitt Analysis (Amplitude against Log(Time))")
plt.xlim(logtsmin, logtsmax)
plt.legend()

plt.show()

示例输出：

Answer 2

使用xlim 和ylim

    plt.plot(np.log(ts), amps, "1", ms=5, label="Schmitt Analysis (Log(Amplitude) 
    against Time)")
    
    
    y1, r, *_ = np.polyfit(amps, ts, 1, full=True)
    f1 = np.poly1d(y1)
    plt.plot(amps, f1(amps), label=f"linear ($\chi^2$ = {r[0]:0.2f})")
    
    plt.xlabel("Log(Time)")
    plt.ylabel("Amplitude")
    plt.title("Schmitt Analysis (Amplitude against Log(Time)")
    plt.xlim(0, 10)
    plt.ylim(0, 10)
    plt.legend()
    
    plt.savefig('A_Schmitt.jpg'

)