Python中简单逻辑回归的错误

Question

import numpy as np
import matplotlib.pyplot as plt


class Logistic_Regression(object):

    def __init__(self, input):
        self.data = np.loadtxt(input, delimiter=',')
        self.x = self.data[:, (0, 1)]
        self.x = np.concatenate((np.ones((100, 1)), self.x), axis=1)
        self.y = self.data[:, 2].reshape((100, 1))
        self.theta = np.zeros((3, 1))
        self.m = 100
        self.hx = self.sigmoid(np.dot(self.x, self.theta))


    def sigmoid(self, para):
        h = np.zeros(para.shape)
        h = 1.0 / (1 + np.exp(-para))
        return h

    def cost_function(self):
        log1 = np.log(self.hx)
        self.hxx = 1.0 - self.hx
        log2 = np.log(self.hxx)
        J = (-1.0) / self.m * (np.dot(self.y.T , log1) + np.dot((1 - self.y).T , log2))
        return J


    def gradient_descent(self):
        a = 0.01
        for i in range(100):
            self.hx = self.sigmoid(np.dot(self.x, self.theta))
            self.theta -= a * np.dot(self.x.T , (self.hx - self.y))
            print self.theta


    def show(self):
        red_orb_x = []
        red_orb_y = []
        blue_orb_x = []
        blue_orb_y = []
        for orb in self.data:
            if orb[2] == 0:
                red_orb_x.append(orb[0])
                red_orb_y.append(orb[1])
            else:
                blue_orb_x.append(orb[0])
                blue_orb_y.append(orb[1])

        plt.plot(red_orb_x, red_orb_y, "ro")
        plt.plot(blue_orb_x, blue_orb_y, "bo")
        #plt.plot(self.x[:, 1], np.dot(self.x, self.theta), "g-")
        plt.show()

    def run(self):
        print "Cost Now is " + str(self.cost_function())
        self.gradient_descent()
        print "After Cost is " + str(self.cost_function())
        self.show()


log_reg = Logistic_Regression("ex2data1.txt")
log_reg.run()

错误是

log_reg.py:26: RuntimeWarning: overflow encountered in exp
  h = 1.0 / (1 + np.exp(-para))
log_reg.py:34: RuntimeWarning: divide by zero encountered in log
  log2 = np.log(hxx)
After Cost is [[ nan]]

我想知道代码中是否有错误？我有一个matlab版本来解决这个数据，我的代码输出第一个成本值，theta是相同的，但我无法得到最终答案。

是梯度下降