很抱歉,如果以前曾问过这个问题,我似乎找不到解决我认为应该是一个简单问题的方法。
我已经找到了一些概率密度数据(5个数据点),这些数据正试图适合合适的分布。
import numpy as np
import matplotlib.pyplot as plt
from scipy.optimize import curve_fit
# Define data to fit to
xdata = [5,18, 26, 30, 40, 50]
ydata = [0.0001, 0.02, 0.36, 0.37, 0.23, 0.02]
# Get midpoints
i = 0
window_size = 2
midx = []
midy = []
while i < len(xdata) - window_size + 1:
this_window = xdata[i : i + window_size]
window_average = sum(this_window) / window_size
midx.append(window_average)
midy.append(ydata[i+1])
i += 1
# Define functional form to fit to
def func(x, s, m, f):
# Log normal function
return (1/(f*x)) * (1/(s * np.sqrt(2 * np.pi))) * np.exp(-((np.log((f*x))-m)**2)/(2*s**2)) #a * np.exp(-b * x) + c
# Execute the fit
popt, pcov = curve_fit(func, midx , midy)
display(popt) # show found
# Plot to see fit
tmpx = np.linspace(0.1, 50, 100)
plt.plot(tmpx, func(tmpx, *popt), 'r-')
plt.step(x=xdata, y=ydata, c='Blue')
plt.scatter(x=midx, y=midy, c='Green')
我明白了:
尽管对于“某种创造性地找到的解决方案”来说这并不可怕,但我还是喜欢正确地做到这一点...
我实际上应该适合于
每对xdata
点之间的函数等于ydata
(即,在x = 5和x = 18.5之间的图形下的表面积应为1.9,在x = 25和x = 18.5之间的表面积应为36.3)
有没有人知道如何执行此操作,最好是使用Scipy之类的库。