sklearn学习笔记（10）——模型的保存和加载

sklearn模型的保存和加载

API：sklearn.externals.joblib

模型保存：joblib.dump(rf, "test.pkl")

模型加载：estimator = joblib.load("test.pkl")

# -*- coding: utf-8 -*-

from sklearn.linear_model import LinearRegression
from sklearn.datasets import load_boston
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
from sklearn.metrics import mean_absolute_error
from sklearn.externals import joblib



def dump_model():
    """
    保存模型
    :return: x_test, y_test, std_y
    """
    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.reshape(-1, 1))
    y_test = std_y.fit_transform(y_test.reshape(-1, 1))

    lr = LinearRegression()
    lr.fit(x_train, y_train)

    # 保存模型到本地文件
    joblib.dump(lr, "test.pkl")

    return x_test, y_test, std_y


def load_model():
    x_test, y_test, std_y = dump_model()

    # 通过模型加载进行预测，使用的模型参数就在本地pkl文件中
    model = joblib.load("test.pkl")
    y_lr_predict = model.predict(x_test)
    result = mean_absolute_error(std_y.inverse_transform(y_test), std_y.inverse_transform(y_lr_predict))
    print("加载模型，lr正规方程求解的均方误差为：", result)


if __name__ == '__main__':
    load_model()