ML笔记-Curve Fitting

机器学习基础学习笔记

多项式拟合$\sin(2\pi x)$,当训练数据点N=10时的情况，多项式的阶数取不同值。

多项式曲线拟合： $y(x,\boldsymbol{w})=w_0+w_1 x + w_2 x^2 + ... + w_M x^M = \sum_{j=0}^{M} w_j x^j$

import matplotlib.pyplot as plt
import numpy as np
from sklearn.pipeline import make_pipeline
from sklearn.preprocessing import PolynomialFeatures
from sklearn.linear_model import LinearRegression

%matplotlib inline
x_ = np.reshape(np.linspace(0, 1, 1000), [-1, 1])
y_ = np.reshape(np.sin(2.*np.pi*x_), [-1, 1])

x = np.linspace(0, 1, 10)
y = np.sin(2.*np.pi*x) + 0.2 * np.random.normal(0, 1, 10)

xtrain = np.reshape(x, [-1, 1])
ytrain = np.reshape(y, [-1, 1])

x_ridge, y_ridge = xtrain, ytrain
x_lasso, y_lasso = xtrain, ytrain

plt.figure()
plt.xlabel(u'x')
plt.ylabel(u'y')
plt.plot(xtrain, ytrain, 'ro')
plt.plot(x_, y_, 'b-')

poly = PolynomialFeatures(10)
x_poly = poly.fit_transform(xtrain)
regressor_poly = LinearRegression()
regressor_poly.fit(x_poly, ytrain)
x__poly = poly.fit_transform(x_)

plt.figure()
plt.xlabel(u'x')
plt.ylabel(u'y')
plt.plot(x, y, 'ro')
plt.plot(x_, y_, 'b-')
plt.plot(x_, regressor_poly.predict(x__poly), 'g-')

# 残差值绘制
yr = regressor_poly.predict(x_poly)
for idx, x