import numpy as np
import matplotlib.pyplot as plt


class LogisticRegress:
    def __init__(self):
        self.theta = theta

    def f(self, x):
        return 1 / (1 + np.exp(-np.dot(x, self.theta)))

    def learning(self, x, y):
        epoch = 5000
        ETA = 1e-3
        count = 0
        accuraries = []
        for _ in range(epoch):
            self.theta = self.theta - ETA * np.dot(self.f(x) - y, x)
            result = self.classify(x) == y
            accurary = len(result[result == True]) / len(result)
            accuraries.append(accurary)
            count += 1
            # print("Times:{}  theta:{}".format(count, self.theta))
        plt.plot(accuraries)
        plt.show()

    def classify(self, x):
        return (self.f(x) >= 0.5).astype(np.int)

    def draw(self):
        x = np.linspace(-2.5, 2.5, 100)
        plt.plot(train_z[train_y == 1, 0], train_z[train_y == 1, 1], 'o')
        plt.plot(train_z[train_y == 0, 0], train_z[train_y == 0, 1], 'x')
        plt.plot(x, -(self.theta[0] + self.theta[1] * x + self.theta[3] * (x ** 2)) / self.theta[2])
        plt.show()


def standardise(x):
    return (x - mu) / sigma


def to_matrix(x):
    x0 = np.ones([x.shape[0], 1])
    return np.hstack([x0, x, x[:, 0, np.newaxis] ** 2])


if __name__ == "__main__":
    train = np.loadtxt("data3.csv", delimiter=',', skiprows=1)
    train_x = train[:, 0:2]
    train_y = train[:, 2]

    theta = np.random.rand(4)
    mu = train_x.mean(axis=0)
    sigma = train_x.std(axis=0)
    train_z = standardise(train_x)
    X = to_matrix(train_z)

    plt.plot(train_z[train_y == 1, 0], train_z[train_y == 1, 1], 'o')
    plt.plot(train_z[train_y == 0, 0], train_z[train_y == 0, 1], 'x')
    plt.show()

    LR = LogisticRegress()
    LR.learning(X, train_y)
    LR.draw()