发布时间:2023-04-19 文章分类:电脑百科 投稿人:樱花 字号: 默认 | | 超大 打印

机器学习实战 | LightGBM建模应用详解

作者:韩信子@ShowMeAI
教程地址:https://www.showmeai.tech/tutorials/41
本文地址:https://www.showmeai.tech/article-detail/205
声明:版权所有,转载请联系平台与作者并注明出处
收藏ShowMeAI查看更多精彩内容

引言

LightGBM是微软开发的boosting集成模型,和XGBoost一样是对GBDT的优化和高效实现,原理有一些相似之处,但它很多方面比XGBoost有着更为优秀的表现。

本篇内容ShowMeAI展开给大家讲解LightGBM的工程应用方法,对于LightGBM原理知识感兴趣的同学,欢迎参考ShowMeAI的另外一篇文章 图解机器学习 | LightGBM模型详解

1.LightGBM安装

LightGBM作为常见的强大Python机器学习工具库,安装也比较简单。

1.1 Python与IDE环境设置

python环境与IDE设置可以参考ShowMeAI文章 图解python | 安装与环境设置 进行设置。

机器学习实战 | LightGBM建模应用详解

1.2 工具库安装

(1) Linux/Mac等系统

这些系统下的XGBoost安装,大家只要基于pip就可以轻松完成了,在命令行端输入命令如下命令即可等待安装完成。

pip install lightgbm

大家也可以选择国内的pip源,以获得更好的安装速度:

pip install -i https://pypi.tuna.tsinghua.edu.cn/simple lightgbm

(2) Windows系统

对于windows系统而言,比较高效便捷的安装方式是:在网址http://www.lfd.uci.edu/~gohlke/pythonlibs/ 中去下载对应版本的的LightGBM安装包,再通过如下命令安装。
pip install lightgbm‑3.3.2‑cp310‑cp310‑win_amd64.whl

2.LightGBM参数手册

在ShowMeAI的前一篇内容 XGBoost工具库建模应用详解 中,我们讲解到了Xgboost的三类参数通用参数,学习目标参数,Booster参数。而LightGBM可调参数更加丰富,包含核心参数,学习控制参数,IO参数,目标参数,度量参数,网络参数,GPU参数,模型参数,这里我常修改的便是核心参数,学习控制参数,度量参数等。下面我们对这些模型参数做展开讲解,更多的细节可以参考LightGBM中文文档。

2.1 参数介绍

(1) 核心参数

(2) 学习控制参数

(3) IO参数

(4) 目标参数

(5) 度量参数

2.2 参数影响与调参建议

以下为总结的核心参数对模型的影响,及与之对应的调参建议。

(1) 对树生长控制

机器学习实战 | LightGBM建模应用详解

(2) 更快的训练速度

机器学习实战 | LightGBM建模应用详解

(3) 更好的模型效果

机器学习实战 | LightGBM建模应用详解

(4) 缓解过拟合问题

机器学习实战 | LightGBM建模应用详解

3.LightGBM内置建模方式

3.1 内置建模方式

LightGBM内置了建模方式,有如下的数据格式与核心训练方法:

下面是官方的一个简单示例,演示了读取libsvm格式数据(成Dataset格式)并指定参数建模的过程。

# coding: utf-8
import json
import lightgbm as lgb
import pandas as pd
from sklearn.metrics import mean_squared_error
# 加载数据集合
print('加载数据...')
df_train = pd.read_csv('./data/regression.train.txt', header=None, sep='\t')
df_test = pd.read_csv('./data/regression.test.txt', header=None, sep='\t')
# 设定训练集和测试集
y_train = df_train[0].values
y_test = df_test[0].values
X_train = df_train.drop(0, axis=1).values
X_test = df_test.drop(0, axis=1).values
# 构建lgb中的Dataset格式
lgb_train = lgb.Dataset(X_train, y_train)
lgb_eval = lgb.Dataset(X_test, y_test, reference=lgb_train)
# 敲定好一组参数
params = {
    'task': 'train',
    'boosting_type': 'gbdt',
    'objective': 'regression',
    'metric': {'l2', 'auc'},
    'num_leaves': 31,
    'learning_rate': 0.05,
    'feature_fraction': 0.9,
    'bagging_fraction': 0.8,
    'bagging_freq': 5,
    'verbose': 0
}
print('开始训练...')
# 训练
gbm = lgb.train(params,
                lgb_train,
                num_boost_round=20,
                valid_sets=lgb_eval,
                early_stopping_rounds=5)
# 保存模型
print('保存模型...')
# 保存模型到文件中
gbm.save_model('model.txt')
print('开始预测...')
# 预测
y_pred = gbm.predict(X_test, num_iteration=gbm.best_iteration)
# 评估
print('预估结果的rmse为:')
print(mean_squared_error(y_test, y_pred) ** 0.5)

机器学习实战 | LightGBM建模应用详解

加载数据...
开始训练...
[1]  valid_0's l2: 0.24288   valid_0's auc: 0.764496
Training until validation scores don't improve for 5 rounds.
[2]  valid_0's l2: 0.239307  valid_0's auc: 0.766173
[3]  valid_0's l2: 0.235559  valid_0's auc: 0.785547
[4]  valid_0's l2: 0.230771  valid_0's auc: 0.797786
[5]  valid_0's l2: 0.226297  valid_0's auc: 0.805155
[6]  valid_0's l2: 0.223692  valid_0's auc: 0.800979
[7]  valid_0's l2: 0.220941  valid_0's auc: 0.806566
[8]  valid_0's l2: 0.217982  valid_0's auc: 0.808566
[9]  valid_0's l2: 0.215351  valid_0's auc: 0.809041
[10] valid_0's l2: 0.213064  valid_0's auc: 0.805953
[11] valid_0's l2: 0.211053  valid_0's auc: 0.804631
[12] valid_0's l2: 0.209336  valid_0's auc: 0.802922
[13] valid_0's l2: 0.207492  valid_0's auc: 0.802011
[14] valid_0's l2: 0.206016  valid_0's auc: 0.80193
Early stopping, best iteration is:
[9]  valid_0's l2: 0.215351  valid_0's auc: 0.809041
保存模型...
开始预测...
预估结果的rmse为:
0.4640593794679212

3.2 设置样本权重

LightGBM的建模非常灵活,它可以支持我们对于每个样本设置不同的权重学习,设置的方式也非常简单,我们需要提供给模型一组权重数组数据,长度和样本数一致。

如下是一个典型的例子,其中binary.trainbinary.test读取后加载为lightgbm.Dataset格式的输入,而在lightgbm.Dataset的构建参数中可以设置样本权重(这个例子中是numpy array的形态)。再基于lightgbm.train接口使用内置建模方式训练。

# coding: utf-8
import json
import lightgbm as lgb
import pandas as pd
import numpy as np
from sklearn.metrics import mean_squared_error
import warnings
warnings.filterwarnings("ignore")
# 加载数据集
print('加载数据...')
df_train = pd.read_csv('./data/binary.train', header=None, sep='\t')
df_test = pd.read_csv('./data/binary.test', header=None, sep='\t')
W_train = pd.read_csv('./data/binary.train.weight', header=None)[0]
W_test = pd.read_csv('./data/binary.test.weight', header=None)[0]
y_train = df_train[0].values
y_test = df_test[0].values
X_train = df_train.drop(0, axis=1).values
X_test = df_test.drop(0, axis=1).values
num_train, num_feature = X_train.shape
# 加载数据的同时加载权重
lgb_train = lgb.Dataset(X_train, y_train,
                        weight=W_train, free_raw_data=False)
lgb_eval = lgb.Dataset(X_test, y_test, reference=lgb_train,
                       weight=W_test, free_raw_data=False)
# 设定参数
params = {
    'boosting_type': 'gbdt',
    'objective': 'binary',
    'metric': 'binary_logloss',
    'num_leaves': 31,
    'learning_rate': 0.05,
    'feature_fraction': 0.9,
    'bagging_fraction': 0.8,
    'bagging_freq': 5,
    'verbose': 0
}
# 产出特征名称
feature_name = ['feature_' + str(col) for col in range(num_feature)]
print('开始训练...')
gbm = lgb.train(params,
                lgb_train,
                num_boost_round=10,
                valid_sets=lgb_train,  # 评估训练集
                feature_name=feature_name,
                categorical_feature=[21])
加载数据...
开始训练...
[1]  training's binary_logloss: 0.68205
[2]  training's binary_logloss: 0.673618
[3]  training's binary_logloss: 0.665891
[4]  training's binary_logloss: 0.656874
[5]  training's binary_logloss: 0.648523
[6]  training's binary_logloss: 0.641874
[7]  training's binary_logloss: 0.636029
[8]  training's binary_logloss: 0.629427
[9]  training's binary_logloss: 0.623354
[10] training's binary_logloss: 0.617593

3.3 模型存储与加载

上述建模过程得到的模型对象,可以通过save_model成员函数进行保存。保存好的模型可以通过lgb.Booster加载回内存,并对测试集进行预测。

具体示例代码如下:

# 查看特征名称
print('完成10轮训练...')
print('第7个特征为:')
print(repr(lgb_train.feature_name[6]))
# 存储模型
gbm.save_model('./model/lgb_model.txt')
# 特征名称
print('特征名称:')
print(gbm.feature_name())
# 特征重要度
print('特征重要度:')
print(list(gbm.feature_importance()))
# 加载模型
print('加载模型用于预测')
bst = lgb.Booster(model_file='./model/lgb_model.txt')
# 预测
y_pred = bst.predict(X_test)
# 在测试集评估效果
print('在测试集上的rmse为:')
print(mean_squared_error(y_test, y_pred) ** 0.5)

机器学习实战 | LightGBM建模应用详解

完成10轮训练...
第7个特征为:
'feature_6'
特征名称:
['feature_0', 'feature_1', 'feature_2', 'feature_3', 'feature_4', 'feature_5', 'feature_6', 'feature_7', 'feature_8', 'feature_9', 'feature_10', 'feature_11', 'feature_12', 'feature_13', 'feature_14', 'feature_15', 'feature_16', 'feature_17', 'feature_18', 'feature_19', 'feature_20', 'feature_21', 'feature_22', 'feature_23', 'feature_24', 'feature_25', 'feature_26', 'feature_27']
特征重要度:
[8, 5, 1, 19, 7, 33, 2, 0, 2, 10, 5, 2, 0, 9, 3, 3, 0, 2, 2, 5, 1, 0, 36, 3, 33, 45, 29, 35]
加载模型用于预测
在测试集上的rmse为:
0.4629245607636925

3.4 继续训练

LightGBM为boosting模型,每一轮训练会增加新的基学习器,LightGBM还支持基于现有模型和参数继续训练,无需每次从头训练。

如下是典型的示例,我们加载已经训练10轮(即10颗树集成)的lgb模型,在此基础上继续训练(在参数层面做了一些改变,调整了学习率,增加了一些bagging等缓解过拟合的处理方法)

# 继续训练
# 从./model/model.txt中加载模型初始化
gbm = lgb.train(params,
                lgb_train,
                num_boost_round=10,
                init_model='./model/lgb_model.txt',
                valid_sets=lgb_eval)
print('以旧模型为初始化,完成第 10-20 轮训练...')
# 在训练的过程中调整超参数
# 比如这里调整的是学习率
gbm = lgb.train(params,
                lgb_train,
                num_boost_round=10,
                init_model=gbm,
                learning_rates=lambda iter: 0.05 * (0.99 ** iter),
                valid_sets=lgb_eval)
print('逐步调整学习率完成第 20-30 轮训练...')
# 调整其他超参数
gbm = lgb.train(params,
                lgb_train,
                num_boost_round=10,
                init_model=gbm,
                valid_sets=lgb_eval,
                callbacks=[lgb.reset_parameter(bagging_fraction=[0.7] * 5 + [0.6] * 5)])
print('逐步调整bagging比率完成第 30-40 轮训练...')

机器学习实战 | LightGBM建模应用详解

[11] valid_0's binary_logloss: 0.616177
[12] valid_0's binary_logloss: 0.611792
[13] valid_0's binary_logloss: 0.607043
[14] valid_0's binary_logloss: 0.602314
[15] valid_0's binary_logloss: 0.598433
[16] valid_0's binary_logloss: 0.595238
[17] valid_0's binary_logloss: 0.592047
[18] valid_0's binary_logloss: 0.588673
[19] valid_0's binary_logloss: 0.586084
[20] valid_0's binary_logloss: 0.584033
以旧模型为初始化,完成第 10-20 轮训练...
[21] valid_0's binary_logloss: 0.616177
[22] valid_0's binary_logloss: 0.611834
[23] valid_0's binary_logloss: 0.607177
[24] valid_0's binary_logloss: 0.602577
[25] valid_0's binary_logloss: 0.59831
[26] valid_0's binary_logloss: 0.595259
[27] valid_0's binary_logloss: 0.592201
[28] valid_0's binary_logloss: 0.589017
[29] valid_0's binary_logloss: 0.586597
[30] valid_0's binary_logloss: 0.584454
逐步调整学习率完成第 20-30 轮训练...
[31] valid_0's binary_logloss: 0.616053
[32] valid_0's binary_logloss: 0.612291
[33] valid_0's binary_logloss: 0.60856
[34] valid_0's binary_logloss: 0.605387
[35] valid_0's binary_logloss: 0.601744
[36] valid_0's binary_logloss: 0.598556
[37] valid_0's binary_logloss: 0.595585
[38] valid_0's binary_logloss: 0.593228
[39] valid_0's binary_logloss: 0.59018
[40] valid_0's binary_logloss: 0.588391
逐步调整bagging比率完成第 30-40 轮训练...

3.5 自定义损失函数

LightGBM支持在训练过程中,自定义损失函数和评估准则,其中损失函数的定义需要返回损失函数一阶和二阶导数的计算方法,评估准则部分需要对数据的label和预估值进行计算。其中损失函数用于训练过程中的树结构学习,而评估准则很多时候是用在验证集上进行效果评估。

# 自定义损失函数需要提供损失函数的一阶和二阶导数形式
def loglikelood(preds, train_data):
    labels = train_data.get_label()
    preds = 1. / (1. + np.exp(-preds))
    grad = preds - labels
    hess = preds * (1. - preds)
    return grad, hess
# 自定义评估函数
def binary_error(preds, train_data):
    labels = train_data.get_label()
    return 'error', np.mean(labels != (preds > 0.5)), False
gbm = lgb.train(params,
                lgb_train,
                num_boost_round=10,
                init_model=gbm,
                fobj=loglikelood,
                feval=binary_error,
                valid_sets=lgb_eval)
print('用自定义的损失函数与评估标准完成第40-50轮...')

机器学习实战 | LightGBM建模应用详解

[41] valid_0's binary_logloss: 0.614429  valid_0's error: 0.268
[42] valid_0's binary_logloss: 0.610689  valid_0's error: 0.26
[43] valid_0's binary_logloss: 0.606267  valid_0's error: 0.264
[44] valid_0's binary_logloss: 0.601949  valid_0's error: 0.258
[45] valid_0's binary_logloss: 0.597271  valid_0's error: 0.266
[46] valid_0's binary_logloss: 0.593971  valid_0's error: 0.276
[47] valid_0's binary_logloss: 0.591427  valid_0's error: 0.278
[48] valid_0's binary_logloss: 0.588301  valid_0's error: 0.284
[49] valid_0's binary_logloss: 0.586562  valid_0's error: 0.288
[50] valid_0's binary_logloss: 0.584056  valid_0's error: 0.288
用自定义的损失函数与评估标准完成第40-50轮...

4.LightGBM预估器形态接口

4.1 SKLearn形态预估器接口

和XGBoost一样,LightGBM也支持用SKLearn中统一的预估器形态接口进行建模,如下为典型的参考案例,对于读取为Dataframe格式的训练集和测试集,可以直接使用LightGBM初始化LGBMRegressor进行fit拟合训练。使用方法与接口,和SKLearn中其他预估器一致。

# coding: utf-8
import lightgbm as lgb
import pandas as pd
from sklearn.metrics import mean_squared_error
from sklearn.model_selection import GridSearchCV
# 加载数据
print('加载数据...')
df_train = pd.read_csv('./data/regression.train.txt', header=None, sep='\t')
df_test = pd.read_csv('./data/regression.test.txt', header=None, sep='\t')
# 取出特征和标签
y_train = df_train[0].values
y_test = df_test[0].values
X_train = df_train.drop(0, axis=1).values
X_test = df_test.drop(0, axis=1).values
print('开始训练...')
# 初始化LGBMRegressor
gbm = lgb.LGBMRegressor(objective='regression',
                        num_leaves=31,
                        learning_rate=0.05,
                        n_estimators=20)
# 使用fit函数拟合
gbm.fit(X_train, y_train,
        eval_set=[(X_test, y_test)],
        eval_metric='l1',
        early_stopping_rounds=5)
# 预测
print('开始预测...')
y_pred = gbm.predict(X_test, num_iteration=gbm.best_iteration_)
# 评估预测结果
print('预测结果的rmse是:')
print(mean_squared_error(y_test, y_pred) ** 0.5)

机器学习实战 | LightGBM建模应用详解

加载数据...
开始训练...
[1]  valid_0's l1: 0.491735
Training until validation scores don't improve for 5 rounds.
[2]  valid_0's l1: 0.486563
[3]  valid_0's l1: 0.481489
[4]  valid_0's l1: 0.476848
[5]  valid_0's l1: 0.47305
[6]  valid_0's l1: 0.469049
[7]  valid_0's l1: 0.465556
[8]  valid_0's l1: 0.462208
[9]  valid_0's l1: 0.458676
[10] valid_0's l1: 0.454998
[11] valid_0's l1: 0.452047
[12] valid_0's l1: 0.449158
[13] valid_0's l1: 0.44608
[14] valid_0's l1: 0.443554
[15] valid_0's l1: 0.440643
[16] valid_0's l1: 0.437687
[17] valid_0's l1: 0.435454
[18] valid_0's l1: 0.433288
[19] valid_0's l1: 0.431297
[20] valid_0's l1: 0.428946
Did not meet early stopping. Best iteration is:
[20] valid_0's l1: 0.428946
开始预测...
预测结果的rmse是:
0.4441153344254208

4.2 网格搜索调参

上面提到LightGBM的预估器接口,整体使用方法和SKLearn中其他预估器一致,所以我们也可以使用SKLearn中的超参数调优方法来进行模型调优。

如下是一个典型的网格搜索交法调优超参数的代码示例,我们会给出候选参数列表字典,通过GridSearchCV进行交叉验证实验评估,选出LightGBM在候选参数中最优的超参数。

# 配合scikit-learn的网格搜索交叉验证选择最优超参数
estimator = lgb.LGBMRegressor(num_leaves=31)
param_grid = {
    'learning_rate': [0.01, 0.1, 1],
    'n_estimators': [20, 40]
}
gbm = GridSearchCV(estimator, param_grid)
gbm.fit(X_train, y_train)
print('用网格搜索找到的最优超参数为:')
print(gbm.best_params_)

机器学习实战 | LightGBM建模应用详解

用网格搜索找到的最优超参数为:
{'learning_rate': 0.1, 'n_estimators': 40}

4.3 绘图解释

LightGBM支持对模型训练进行可视化呈现与解释,包括对于训练过程中的损失函数取值与评估准则结果的可视化、训练完成后特征重要度的排序与可视化、基学习器(比如决策树)的可视化。

以下为参考代码:

# coding: utf-8
import lightgbm as lgb
import pandas as pd
try:
    import matplotlib.pyplot as plt
except ImportError:
    raise ImportError('You need to install matplotlib for plotting.')
# 加载数据集
print('加载数据...')
df_train = pd.read_csv('./data/regression.train.txt', header=None, sep='\t')
df_test = pd.read_csv('./data/regression.test.txt', header=None, sep='\t')
# 取出特征和标签
y_train = df_train[0].values
y_test = df_test[0].values
X_train = df_train.drop(0, axis=1).values
X_test = df_test.drop(0, axis=1).values
# 构建lgb中的Dataset数据格式
lgb_train = lgb.Dataset(X_train, y_train)
lgb_test = lgb.Dataset(X_test, y_test, reference=lgb_train)
# 设定参数
params = {
    'num_leaves': 5,
    'metric': ('l1', 'l2'),
    'verbose': 0
}
evals_result = {}  # to record eval results for plotting
print('开始训练...')
# 训练
gbm = lgb.train(params,
                lgb_train,
                num_boost_round=100,
                valid_sets=[lgb_train, lgb_test],
                feature_name=['f' + str(i + 1) for i in range(28)],
                categorical_feature=[21],
                evals_result=evals_result,
                verbose_eval=10)
print('在训练过程中绘图...')
ax = lgb.plot_metric(evals_result, metric='l1')
plt.show()
print('画出特征重要度...')
ax = lgb.plot_importance(gbm, max_num_features=10)
plt.show()
print('画出第84颗树...')
ax = lgb.plot_tree(gbm, tree_index=83, figsize=(20, 8), show_info=['split_gain'])
plt.show()
#print('用graphviz画出第84颗树...')
#graph = lgb.create_tree_digraph(gbm, tree_index=83, name='Tree84')
#graph.render(view=True)

机器学习实战 | LightGBM建模应用详解

加载数据...
开始训练...
[10] training's l2: 0.217995 training's l1: 0.457448 valid_1's l2: 0.21641   valid_1's l1: 0.456464
[20] training's l2: 0.205099 training's l1: 0.436869 valid_1's l2: 0.201616  valid_1's l1: 0.434057
[30] training's l2: 0.197421 training's l1: 0.421302 valid_1's l2: 0.192514  valid_1's l1: 0.417019
[40] training's l2: 0.192856 training's l1: 0.411107 valid_1's l2: 0.187258  valid_1's l1: 0.406303
[50] training's l2: 0.189593 training's l1: 0.403695 valid_1's l2: 0.183688  valid_1's l1: 0.398997
[60] training's l2: 0.187043 training's l1: 0.398704 valid_1's l2: 0.181009  valid_1's l1: 0.393977
[70] training's l2: 0.184982 training's l1: 0.394876 valid_1's l2: 0.178803  valid_1's l1: 0.389805
[80] training's l2: 0.1828   training's l1: 0.391147 valid_1's l2: 0.176799  valid_1's l1: 0.386476
[90] training's l2: 0.180817 training's l1: 0.388101 valid_1's l2: 0.175775  valid_1's l1: 0.384404
[100]   training's l2: 0.179171 training's l1: 0.385174 valid_1's l2: 0.175321  valid_1's l1: 0.382929

机器学习实战 | LightGBM建模应用详解

机器学习实战 | LightGBM建模应用详解

参考资料

ShowMeAI系列教程推荐

相关文章推荐

机器学习实战 | LightGBM建模应用详解