机器学习之回归算法1

机器学习之回归算法1

线性回归

定义

损失函数

求解

正常求解

梯度下降


比较

过拟合和欠拟合

岭回归

代码实现

from sklearn.datasets import load_boston
from sklearn.linear_model import LinearRegression, SGDRegressor, Ridge
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
from sklearn.metrics import mean_absolute_error

def myliner():
    lb = load_boston()
    x_train, x_test, y_train, y_test = train_test_split(lb.data, lb.target, test_size=0.25)
    std_x = StandardScaler()
    x_train = std_x.fit_transform(x_train)
    x_test = std_x.transform(x_test)
    std_y = StandardScaler()
    y_train = std_y.fit_transform(y_train)
    # y_train = std_y.fit_transform(y_train.reshape[-1, 1]) 版本不同，有的要求二维
    y_test = std_y.transform(y_test)
    lr = LinearRegression()
    lr.fit(x_train, x_test)
    lr.fit(x_train, y_train)
    print(lr.coef_)
    y_predict_lr = std_y.inverse_transform(lr.predict(x_test))
    print('预测价格', y_predict_lr)
    print('均方误差', mean_absolute_error(std_y.inverse_transform(y_test), y_predict_lr))
    # 梯度下降
    sgd = SGDRegressor()
    sgd.fit(x_train, x_test)
    print(sgd.coef_)
    y_predict_sgd = std_y.inverse_transform(sgd.predict(x_test))
    print('预测价格', y_predict_sgd)
    print('均方误差', mean_absolute_error(std_y.inverse_transform(y_test), y_predict_sgd))
    # 岭回归
    rd = Ridge(alpha=1.0)
    rd.fit(x_train, x_test)
    print(rd.coef_)
    y_predict_rd = std_y.inverse_transform(rd.predict(x_test))
    print('预测价格', y_predict_rd)
    print('均方误差', mean_absolute_error(std_y.inverse_transform(y_test), y_predict_rd))
    return

myliner()