目錄

一、基礎(chǔ)理論

API

二、嶺回歸：預(yù)測波士頓房價(jià)

總代碼

---

一、基礎(chǔ)理論

嶺回歸：帶有L2正則化的線性回歸。（為了解決過擬合）

對病態(tài)數(shù)據(jù)的擬合要強(qiáng)于最小二乘法

API

sklearn.linear_model.Ridge

?

?（橫坐標(biāo)：正則化力度；? ? ? ? 縱坐標(biāo)：權(quán)重系數(shù)）

二、嶺回歸：預(yù)測波士頓房價(jià)

# 3、嶺回歸
def Linear3():# 1、獲取數(shù)據(jù)集boston = load_boston()# print(boston)# 2、劃分?jǐn)?shù)據(jù)集train_data, test_data, train_target, test_target = train_test_split(boston.data, boston.target, random_state=22)# print(train_data)# 3、標(biāo)準(zhǔn)化transfer = StandardScaler()train_data = transfer.fit_transform(train_data)test_data = transfer.transform(test_data)# print(train_data)# 4、創(chuàng)建預(yù)估器estimator = Ridge()                     # 嶺回歸estimator.fit(train_data, train_target) # 訓(xùn)練# 得到模型print('嶺回歸 權(quán)重系數(shù)：', estimator.coef_)print('嶺回歸 偏置：', estimator.intercept_)# 5、模型評估predict = estimator.predict(test_data)error = mean_squared_error(test_target, predict)print('嶺回歸 均方差：', error)

總代碼

# 線性回歸：波士頓房價(jià)預(yù)測（正規(guī)方程，梯度下降，嶺回歸）
from sklearn.datasets import load_boston
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
from sklearn.linear_model import LinearRegression,SGDRegressor,Ridge
from sklearn.metrics import mean_squared_error# 1、正規(guī)方程優(yōu)化
def Linear1():# 1、獲取數(shù)據(jù)集boston = load_boston()# print(boston)# 2、劃分?jǐn)?shù)據(jù)集train_data, test_data, train_target, test_target = train_test_split(boston.data, boston.target, random_state=22)# print(train_data)# 3、標(biāo)準(zhǔn)化transfer = StandardScaler()train_data = transfer.fit_transform(train_data)test_data = transfer.transform(test_data)# print(train_data)# 4、創(chuàng)建預(yù)估器，得到模型estimator = LinearRegression()              #正規(guī)方程優(yōu)化estimator.fit(train_data, train_target)     #訓(xùn)練# showprint('正規(guī)方程 權(quán)重系數(shù)：', estimator.coef_)print('正規(guī)方程 偏置：', estimator.intercept_)# 5、模型評估predict = estimator.predict(test_data)error = mean_squared_error(test_target, predict)print('正規(guī)方程 均方差：', error)# 2、梯度下降優(yōu)化
def Linear2():# 1、獲取數(shù)據(jù)集boston = load_boston()# print(boston)# 2、劃分?jǐn)?shù)據(jù)集train_data, test_data, train_target, test_target = train_test_split(boston.data, boston.target, random_state=22)# print(train_data)# 3、標(biāo)準(zhǔn)化transfer = StandardScaler()train_data = transfer.fit_transform(train_data)test_data = transfer.transform(test_data)# print(train_data)# 4、創(chuàng)建預(yù)估器estimator = SGDRegressor()                  #梯度下降優(yōu)化estimator.fit(train_data, train_target)     #訓(xùn)練# 得到模型print('梯度下降 權(quán)重系數(shù)：', estimator.coef_)print('梯度下降 偏置：', estimator.intercept_)# 5、模型評估predict = estimator.predict(test_data)error = mean_squared_error(test_target, predict)print('梯度下降 均方差：', error)# 3、嶺回歸
def Linear3():# 1、獲取數(shù)據(jù)集boston = load_boston()# print(boston)# 2、劃分?jǐn)?shù)據(jù)集train_data, test_data, train_target, test_target = train_test_split(boston.data, boston.target, random_state=22)# print(train_data)# 3、標(biāo)準(zhǔn)化transfer = StandardScaler()train_data = transfer.fit_transform(train_data)test_data = transfer.transform(test_data)# print(train_data)# 4、創(chuàng)建預(yù)估器estimator = Ridge()                     # 嶺回歸estimator.fit(train_data, train_target) # 訓(xùn)練# 得到模型print('嶺回歸 權(quán)重系數(shù)：', estimator.coef_)print('嶺回歸 偏置：', estimator.intercept_)# 5、模型評估predict = estimator.predict(test_data)error = mean_squared_error(test_target, predict)print('嶺回歸 均方差：', error)if __name__ == '__main__':Linear1()           # 正規(guī)方程Linear2()           # 梯度下降Linear3()           # 嶺回歸

總結(jié)

以上是生活随笔為你收集整理的机器学习（13）岭回归（线性回归的改进）的全部內(nèi)容，希望文章能夠幫你解決所遇到的問題。

如果覺得生活随笔網(wǎng)站內(nèi)容還不錯(cuò)，歡迎將生活随笔推薦給好友。