线性回归算法梳理 task2

1.生成数据

x = np.array([0,1,2,3,4,5,6,7,8,9,10])
y = np.array([12,14,18,19,21,23,24,26,27,29,31])#显示数据

plt.scatter(x,y)
plt.show()

2.算术法求解

#对数据集进行3：7的拆分
x_train = x[:7].reshape(-1,1)
x_test = x[7:].reshape(-1,1)
y_train = y[:7]
y_test = y[7:]

然后利用公式对系数和截距进行计算
$$b1=\frac{{\sum }_{i=1}^n(x^{(i)}-\overline{x})(y^{(i)}-\overline{y})}{{\sum }_{i=1}^n(x^{(i)}-\overline{x}{)}^2}$$
$$b0=\overline{y}-b1\ast \overline{x}$$

num = 0
d = 0
n = len(x_train)
for i in range(0,len(x_train)):
    num += (x_train[i]-np.mean(x_train))*(y_train[i]-np.mean(y_train)) #分子
    d += (x_train[i]-np.mean(x_train))**2 #分母
b1 = 1.0*num / float(d)
b0 = np.mean(y_train) - b1*np.mean(x_train)

得到b0,b1的结果分别为

(array([2.03571429]), array([12.60714286]))

将直线绘制出来

x_ = np.linspace(0,10,100)
y_ = b1*x_ + b0
plt.scatter(x,y)
plt.plot(x_,y_,c='r')
plt.show()

3.向量法求解

regression = linear_model.LinearRegression()#创建回归模型
regression.fit(x_train,y_train)
b0,b1 = regression.intercept_,regression.coef_

得到b1,b0的值分别为

(12.60714285714286, array([2.03571429]))

再将直线绘制出来

4.算术法向量法的性能比较

通过计算两者的SSE进行比较

SSE = 0
pre = b0 + b1*x_test
for i in range(0,4):
    SSE += (y_test[i]-pre[i])**2
print(SSE)

得到两者的SSE都大约为

[11.89540816]