LGBM使用贝叶斯调参

贝叶斯调参的相关知识与代码格式请参考：https://blog.csdn.net/weixin_35757704/article/details/118416689

1. 构造LGBM模型

       model = LGBMRegressor(
           num_leaves=31,
           learning_rate=learning_rate,
           n_estimators=int(n_estimators),
           silent=True,
           n_jobs=5, random_state=0, verbose=-1,
           min_child_samples=int(min_child_samples),
           feature_fraction=feature_fraction,
           bagging_fraction=bagging_fraction,
       )
       model.fit(x_train, y_train, eval_set=[(x_test, y_test)], verbose=-1)
   

2. 使用贝叶斯模板
3. 训练模型，并且将每一次的结果都写入文件中，防止程序意外终止。

完整代码

from lightgbm import LGBMRegressor
from sklearn.model_selection import train_test_split
from sklearn.datasets import make_regression
import numpy as np
from bayes_opt import BayesianOptimization
from sklearn.metrics import mean_squared_error


def train_model(learning_rate, n_estimators, min_child_samples, feature_fraction, bagging_fraction):
    # 模型训练
    try:
        model = LGBMRegressor(
            num_leaves=31,
            learning_rate=learning_rate,
            n_estimators=int(n_estimators),
            silent=True,
            n_jobs=5, random_state=0, verbose=-1,
            min_child_samples=int(min_child_samples),
            feature_fraction=feature_fraction,
            bagging_fraction=bagging_fraction,
        )
        model.fit(x_train, y_train, eval_set=[(x_test, y_test)], verbose=-1)
        score = - mean_squared_error(y_test, model.predict(x_test, num_iteration=model.best_iteration_))
        with open('lgbm_param.txt', 'a') as file:
            file.write("learning_rate:" + str(learning_rate) + " n_estimators:" + str(
                int(n_estimators)) + " min_child_samples:" + str(int(min_child_samples)) + " score:" + str(
                score) + '\n')
        return score
    except Exception as e:
        return -1000000


if __name__ == '__main__':
    x, y = make_regression(n_samples=1000, n_features=5)
    x_train, x_test, y_train, y_test = train_test_split(x, y, test_size=0.2)

    pbounds = {'learning_rate': (0.0001, 0.1),
               'n_estimators': (100, 1000),
               'min_child_samples': (18, 40),
               'feature_fraction': (0.5, 0.9999999999),
               "bagging_fraction": (0.5, 0.99999999999),
               }
    optimizer = BayesianOptimization(
        f=train_model,  # 黑盒目标函数
        pbounds=pbounds,  # 取值空间
        verbose=2,  # verbose = 2 时打印全部，verbose = 1 时打印运行中发现的最大值，verbose = 0 将什么都不打印
        random_state=1,
    )
    optimizer.maximize(  # 运行
        init_points=10,  # 随机搜索的步数
        n_iter=5000,  # 执行贝叶斯优化迭代次数
    )
    with open('lgbm_param.txt', 'a') as file:
        file.write("optimizer_params: " + str(optimizer.max['params']) + " optimizer_target: " + str(
            optimizer.max['target']) + '\n')