数据拟合

题目地址：http://www.qlcoder.com/task/763b

"""
有一个1000行的文本文件，文件的每一行有2个浮点数，代表一个点的坐标(x,y)。（所有点都在第一象限）

请找出一条直线 Y=a*X+b，使文件中所有点和该直线的距离的和最短。保留两位小数

思路：f(x) = mx + k;
需要注意的地方，迭代的次数和学习速率的设置
"""

import numpy as np
import matplotlib.pyplot as plt



def compute_error(m,k,points):
    totalError = 0
    for i in range(len(points)):
        x = points[i,0]
        y = points[i,1]
        totalError += (y - (m * x + k))**2
    return totalError / float(len(points))

def step_gradient(m_current, k_current, points, learningRate):
    m_gradient = 0.0
    k_gradient = 0.0
    N = float(len(points))
    for i in range(len(points)):
        x = points[i, 0]
        y = points[i, 1]
        m_gradient += - (2/N)*x*(y- (m_current*x + k_current))
        k_gradient += - (2/N)*(y - (m_current*x + k_current))
    new_m = m_current - learningRate*m_gradient
    new_k = k_current - learningRate*k_gradient
    return [new_m,new_k]

def gradient_descent_runner(points, starting_m, starting_k, learning_rate, num_iterations):
    m = starting_m
    k = starting_k
    for i in range(num_iterations):
        m,k = step_gradient(m, k, np.array(points), learning_rate)
    return [m,k]
def run():
    points = np.genfromtxt("data.txt", delimiter=' ')
    learning_rate = 0.0001
    init_m = 0
    init_k = 0
    num_iterations = 2000
    print("Starting gradient descent at b = {0}, m = {1}, error = {2}".format(init_m, init_k, compute_error(init_m, init_k, points)))
    print("Running...")
    [m, k] = gradient_descent_runner(points, init_m, init_k, learning_rate, num_iterations)
    print("After {0} iterations b = {1}, m = {2}, error = {3}".format(num_iterations, m, k, compute_error(m, k, points)))
    print('----------------------')
    print("{0}-{1}".format(m,k))
def draw():
    points = np.array(np.genfromtxt("data.txt", delimiter=' '))
    x = points[::,0]
    y = points[::,1]
    plt.scatter(x,y)
    plt.show()

if __name__ == "__main__":
    run()
    # draw()