如何将下面的代码写成一个class：def oprobit(theta): beta=theta[0] BX = x*beta mu1=theta[1] mu2=theta[2] mu3=theta[3] mu4=theta[4] output=np.copy(y) part0=(np.log(stats.norm.cdf(mu1-BX[y==0])-stats.norm.cdf(-np.inf-BX[y==0]))) part1=(np.log(stats.norm.cdf(mu2-BX[y==1])-stats.norm.cdf(mu1-BX[y==1]))) part2=(np.log(stats.norm.cdf(mu3-BX[y==2])-stats.norm.cdf(mu2-BX[y==2]))) part3=(np.log(stats.norm.cdf(mu4-BX[y==3])-stats.norm.cdf(mu3-BX[y==3]))) part4=(np.log(stats.norm.cdf(np.inf-BX[y==4])-stats.norm.cdf(mu4-BX[y==4]))) return -(part0.sum()+part1.sum()+part2.sum()+part3.sum()+part4.sum()) minimize(oprobit, x0=np.array([0.8,4,5,6,7]))

可以将代码写成如下的class：

import numpy as np
from scipy import stats
from scipy.optimize import minimize

class Oprobit:
    def __init__(self, x, y):
        self.x = x
        self.y = y
    
    def __call__(self, theta):
        beta = theta[0]
        BX = self.x * beta
        mu1 = theta[1]
        mu2 = theta[2]
        mu3 = theta[3]
        mu4 = theta[4]
        output = np.copy(self.y)
        part0 = (np.log(stats.norm.cdf(mu1 - BX[self.y == 0]) - stats.norm.cdf(-np.inf - BX[self.y == 0])))
        part1 = (np.log(stats.norm.cdf(mu2 - BX[self.y == 1]) - stats.norm.cdf(mu1 - BX[self.y == 1])))
        part2 = (np.log(stats.norm.cdf(mu3 - BX[self.y == 2]) - stats.norm.cdf(mu2 - BX[self.y == 2])))
        part3 = (np.log(stats.norm.cdf(mu4 - BX[self.y == 3]) - stats.norm.cdf(mu3 - BX[self.y == 3])))
        part4 = (np.log(stats.norm.cdf(np.inf - BX[self.y == 4]) - stats.norm.cdf(mu4 - BX[self.y == 4])))
        return -(part0.sum() + part1.sum() + part2.sum() + part3.sum() + part4.sum())

然后可以通过如下方式调用：

model = Oprobit(x, y)
result = minimize(model, x0=np.array([0.8, 4, 5, 6, 7]))