PyTorch实战课程（2）——回归问题

 

 

import numpy as np


# y = wx + b


# 计算avg_loss
def compute_error_for_line_given_points(b, w, points):
    totalError = 0
    for i in range(0, len(points)):  # range()左闭右开
        x = points[i, 0]
        y = points[i, 1]
        totalError += (y - (w * x + b)) ** 2
    return totalError / float(len(points))


# 进行梯度下降
def step_gradient(b_current, w_current, points, learningRate):
    b_gradient = 0
    w_gradient = 0
    N = float(len(points))
    for i in range(0, len(points)):
        x = points[i, 0]
        y = points[i, 1]
        b_gradient += 2 / N * ((w_current * x) + b_current - y)
        w_gradient += 2 / N * ((w_current * x) + b_current - y) * x
    new_b = b_current - learningRate * b_gradient
    new_w = w_current - learningRate * w_gradient
    return [new_b, new_w]


# 运行梯度下降
def gradient_descent_runner(points, starting_b, starting_w, learning_rate, num_iterations):
    b = starting_b
    w = starting_w
    for i in range(0, len(points)):
        b, w = step_gradient(b, w, np.array(points), learning_rate)
    return [b, w]


def run():
    points = np.genfromtxt("data.csv", delimiter=",")
    learning_rate = 0.0001
    initial_b = 0
    initial_w = 0
    num_iterations = 100
    print("Starting gradient descent at b = {0} , w = {1}, error = {2}"
          .format(initial_b, initial_w,
                  compute_error_for_line_given_points(initial_b, initial_w, points)))
    print("Running...")
    [b, w] = gradient_descent_runner(points, initial_b, initial_w, learning_rate, num_iterations)
    print("After {0} iterations b = {1}, w = {2}, error = {3}"
          .format(num_iterations, b, w,
                  compute_error_for_line_given_points(b, w, points)))


run()

运行结果：

 data.csv数据：

32.50235	31.70701
53.4268	68.7776
61.53036	62.56238
47.47564	71.54663
59.81321	87.23093
55.14219	78.21152
52.2118	79.64197
39.29957	59.17149
48.10504	75.33124
52.55001	71.30088
45.41973	55.16568
54.35163	82.47885
44.16405	62.00892
58.16847	75.39287
56.72721	81.43619
48.95589	60.7236
44.6872	82.8925
60.29733	97.3799
45.61864	48.84715
38.81682	56.87721
66.18982	83.87856
65.41605	118.5912
47.48121	57.25182
41.57564	51.39174
51.84519	75.38065
59.37082	74.76556
57.31	95.45505
63.61556	95.22937
46.73762	79.05241
50.55676	83.43207
52.224	63.35879
35.56783	41.41289
42.43648	76.61734
58.16454	96.76957
57.50445	74.08413
45.44053	66.58814
61.89622	77.76848
33.09383	50.71959
36.43601	62.12457
37.67565	60.81025
44.55561	52.68298
43.31828	58.56982
50.07315	82.90598
43.87061	61.42471
62.99748	115.2442
32.66904	45.57059
40.1669	54.08405
53.57508	87.99445
33.86421	52.72549
64.70714	93.57612
38.11982	80.16628
44.50254	65.10171
40.59954	65.5623
41.72068	65.28089
51.08863	73.43464
55.0781	71.13973
41.37773	79.10283
62.4947	86.52054
49.20389	84.7427
41.10269	59.35885
41.18202	61.68404
50.18639	69.8476
52.37845	86.09829
50.13549	59.10884
33.64471	69.89968
39.5579	44.86249
56.13039	85.49807
57.36205	95.53669
60.26921	70.25193
35.67809	52.72173
31.58812	50.39267
53.66093	63.6424
46.68223	72.24725
43.10782	57.81251
70.34608	104.2571
44.49286	86.64202
57.50453	91.48678
36.93008	55.23166
55.80573	79.55044
38.95477	44.84712
56.90121	80.20752
56.8689	83.14275
34.33312	55.72349
59.04974	77.63418
57.78822	99.05141
54.28233	79.12065
51.08872	69.5889
50.28284	69.5105
44.21174	73.68756
38.00549	61.3669
32.94048	67.17066
53.69164	85.6682
68.76573	114.8539
46.23097	90.12357
68.31936	97.91982
50.03017	81.53699
49.23977	72.11183
50.03958	85.23201
48.14986	66.22496
25.12848	53.45439