Pipelines + GridSearch

最新推荐文章于 2024-12-26 03:04:26 发布

mym_74

最新推荐文章于 2024-12-26 03:04:26 发布

阅读量135

点赞数

CC 4.0 BY-SA版权

分类专栏：机器学习算法

本文链接：https://blog.csdn.net/mym_74/article/details/102887298

机器学习算法专栏收录该内容

8 篇文章

订阅专栏

本文详细介绍了使用Python的scikit-learn和LightGBM库进行机器学习模型优化的过程。通过GridSearchCV进行参数调优，结合Imputer处理缺失值，SelectKBest特征选择，以及LGBMRegressor作为预测模型，实现对连续型数据的有效预测，并评估模型性能。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

加载库与数据

直接给出示例

import numpy as np
import pandas as pd
import lightgbm as lgb
from lightgbm import LGBMRegressor
from sklearn.model_selection import GridSearchCV
from sklearn.feature_selection import SelectKBest, VarianceThreshold
from sklearn.pipeline import Pipeline, FeatureUnion
from sklearn.preprocessing import Imputer, PolynomialFeatures, StandardScaler, OneHotEncoder, MinMaxScaler
from sklearn.decomposition import PCA
from sklearn.ensemble import RandomForestRegressor
import os
import time

import warnings
warnings.filterwarnings("ignore")

def MAE(y, ypred):  #定义损失函数


    import numpy as np

    return np.sum([abs(y[i]-ypred[i]) for i in range(len(y))]) / len(y)

pipeline = Pipeline(
                    [('imp', Imputer(missing_values='NaN', axis=0)),
                     ('feat_select', SelectKBest()),
                     ('lgbm', LGBMRegressor())
                     
])

parameters = {}
parameters['imp__strategy'] = ['mean', 'median', 'most_frequent']
parameters['feat_select__k'] = [5, 10]

CV = GridSearchCV(pipeline, parameters, scoring = 'mean_absolute_error', n_jobs= 1)
CV.fit(x_train_cont, y_train)   

print('Best score and parameter combination = ')

print(CV.best_score_)    
print(CV.best_params_)    

y_pred = CV.predict(x_valid_cont)
print('MAE on validation set: %s' % (round(MAE(y_valid, y_pred), 5)))