机器学习——梯度下降算法的线性模型的优化

x_data = [338., 333., 328., 207., 226., 25., 179., 60., 208., 606.]
y_data = [640., 633., 619., 393., 428., 27., 193., 66., 226., 1591.]

[4]:

import numpy as np
import matplotlib.pyplot as plt
from pylab import mpl

[5]:

mpl.rcParams['font.sans-serif'] = ['FangSong']
mpl.rcParams['axes.unicode_minus'] = False

[6]:

x = np.arange(-200, -100, 1)
y = np.arange(-5, 5, 0.1)
z = np.zeros((len(x), len(y)))
X, Y = np.meshgrid(x, y)
for i in range(len(x)):
    for j in range(len(y)):
        b = x[i]
        w = y[j]
        z[j][i] = 0
        for n in range(len(x_data)):
            z[j][i] = z[j][i] + (y_data[n] - b - w * x_data[n]) ** 2
        z[j][i] = z[j][i] / len(x_data)

[13]:

plt.contourf(x, y, z, 50, alpha=0.5, cmap=plt.get_cmap('jet'))

[13]:

<matplotlib.contour.QuadContourSet at 0x14f812f5910>

[8]:

b = -120
w = -4
lr = 1
iteration = 100000

[9]:

b_history = [b]
w_history = [w]

[10]:

history_b_grad = 0.0
history_w_grad = 0.0

[11]:

for i in range(iteration):
    w_grad=0.0
    b_grad=0.0
    for i in range(len(x_data)):
        w_grad+=2*(b+w*x_data[i]-y_data[i])*x_data[i]
        b_grad+=2*(b+w*x_data[i]-y_data[i])
    history_b_grad+=b_grad**2
    history_w_grad+=w_grad**2
    w=w-lr/np.sqrt(history_w_grad)*w_grad
    b=b-lr/np.sqrt(history_b_grad)*b_grad
    w_history.append(w)
    b_history.append(b)

[12]:

plt.contourf(x, y, z, 50, alpha=0.5, cmap=plt.get_cmap('jet'))  # 填充等高线
plt.plot([-188.4], [2.67], 'x', ms=12, mew=3, color="orange")
plt.plot(b_history, w_history, 'o-', ms=3, lw=1.5, color='black')
plt.xlim(-200, -100)
plt.ylim(-5, 5)
plt.xlabel(r'$b$')
plt.ylabel(r'$w$')
plt.title("线性回归")
plt.show()

完整代码

x_data = [338., 333., 328., 207., 226., 25., 179., 60., 208., 606.]
y_data = [640., 633., 619., 393., 428., 27., 193., 66., 226., 1591.]
# %%
import numpy as np
import matplotlib.pyplot as plt
from pylab import mpl

mpl.rcParams['font.sans-serif'] = ['FangSong']
mpl.rcParams['axes.unicode_minus'] = False
# %%
x = np.arange(-200, -100, 1)
y = np.arange(-5, 5, 0.1)
z = np.zeros((len(x), len(y)))
X, Y = np.meshgrid(x, y)
for i in range(len(x)):
    for j in range(len(y)):
        b = x[i]
        w = y[j]
        z[j][i] = 0
        for n in range(len(x_data)):
            z[j][i] = z[j][i] + (y_data[n] - b - w * x_data[n]) ** 2
        z[j][i] = z[j][i] / len(x_data)
# %%
plt.contourf(x, y, z, 50, alpha=0.5, cmap=plt.get_cmap('jet'))
# %%
# 初始化w,b值
b = -120
w = -4
lr = 1
iteration = 100000

# 记录w,b的变化过程
b_history = [b]
w_history = [w]

# 梯度下降
history_b_grad = 0.0
history_w_grad = 0.0
for i in range(iteration):
    w_grad=0.0
    b_grad=0.0
    for i in range(len(x_data)):
        w_grad+=2*(b+w*x_data[i]-y_data[i])*x_data[i]
        b_grad+=2*(b+w*x_data[i]-y_data[i])
    history_b_grad+=b_grad**2
    history_w_grad+=w_grad**2
    w=w-lr/np.sqrt(history_w_grad)*w_grad
    b=b-lr/np.sqrt(history_b_grad)*b_grad
    w_history.append(w)
    b_history.append(b)

# 记录每次梯度更新后的参数值


# 可视化参数变化过程
plt.contourf(x, y, z, 50, alpha=0.5, cmap=plt.get_cmap('jet'))  # 填充等高线
plt.plot([-188.4], [2.67], 'x', ms=12, mew=3, color="orange")
plt.plot(b_history, w_history, 'o-', ms=3, lw=1.5, color='black')
plt.xlim(-200, -100)
plt.ylim(-5, 5)
plt.xlabel(r'$b$')
plt.ylabel(r'$w$')
plt.title("线性回归")
plt.show()