前言

16年底，微软DMTK(分布式机器学习工具包)团队在GitHub上开源了性能超越其他boosting工具的LightGBM，在三天之内GitHub上被star了1000次，fork了200次，可见LightGBM的火爆程度。

在前面我们也说过，GBDT (Gradient Boosting Decision Tree)是机器学习中一个长盛不衰的模型，其主要思想是利用弱分类器（决策树）迭代训练以得到最优模型，该模型具有训练效果好、不易过拟合等优点。GBDT在工业界应用广泛，通常被用于点击率预测，搜索排序等任务。GBDT也是各种数据挖掘竞赛的致命武器，据统计Kaggle上的比赛有一半以上的冠军方案都是基于GBDT。Xgboost是GBDT的集大成者，但是LightGBM的出现挑战了Xgboost在“江湖”上的地位

LightGBM （Light Gradient Boosting Machine）(官方github，英文官方文档，中文官方文档)是一个实现GBDT算法的轻量级框架，支持高效率的并行训练，并且具有以下优点：

• 更快的训练效率
• 低内存使用
• 更高的准确率
• 支持并行化学习
• 可处理大规模数据

Python代码实现

分类模型

训练模型

import lightgbm as gbm
from sklearn.datasets import load_iris
from matplotlib import pyplot as plt
from sklearn.model_selection import train_test_split

# read in the iris data
iris = load_iris()

X = iris.data
y = iris.target

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=0)

# 训练模型
model = gbm.LGBMClassifier(max_depth=5, learning_rate=0.1, n_estimators=160, silent=True, objective='multi:softmax')
model.fit(X_train, y_train)

# 对测试集进行预测
ans = model.predict(X_test)

model.score(X_test,y_test)

绘制模型重要性

plt.figure()
gbm.plot_importance(model)
plt.show()

绘制模型

plt.figure()
gbm.plot_tree(model)
plt.show()

回归模型

训练模型

import lightgbm as gbm
from sklearn.datasets import load_iris
from matplotlib import pyplot as plt
from sklearn.model_selection import train_test_split

# read in the iris data
iris = load_iris()

X = iris.data[:,:3]
y = iris.data[:,3]

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=0)

# 训练模型
model = gbm.LGBMRegressor(max_depth=5, learning_rate=0.1, n_estimators=160, silent=True)
model.fit(X_train, y_train)

# 对测试集进行预测
ans = model.predict(X_test)

model.score(X_test,y_test)

绘制模型重要性

plt.figure()
gbm.plot_importance(model)
plt.show()

绘制模型

plt.figure()
gbm.plot_tree(model)
plt.show()

参考资料

• https://blog.csdn.net/anshuai_aw1/article/details/83659932

---

import lightgbm as gbm
from sklearn.datasets import load_iris
from matplotlib import pyplot as plt
from sklearn.model_selection import train_test_split
import pandas as pd
# read in the iris data
iris = load_iris()

X = pd.DataFrame(iris.data,columns=iris.feature_names)
y = iris.target

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=0)

# 训练模型
model = gbm.LGBMClassifier(max_depth=5, learning_rate=0.1, n_estimators=160, silent=True, objective='multi:softmax')
model.fit(X_train, y_train)

# 对测试集进行预测
ans = model.predict(X_test)

model.score(X_test,y_test)

0.9666666666666667

plt.figure()
gbm.plot_importance(model)
plt.show()

<Figure size 640x480 with 0 Axes>

plt.figure()
gbm.plot_tree(model)
plt.show()

<Figure size 640x480 with 0 Axes>

import lightgbm as gbm
from sklearn.datasets import load_iris
from matplotlib import pyplot as plt
from sklearn.model_selection import train_test_split

# read in the iris data
iris = load_iris()

X = pd.DataFrame(iris.data[:,:3],columns=iris.feature_names[:3])
y = iris.data[:,3]

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=0)

# 训练模型
model = gbm.LGBMRegressor(max_depth=5, learning_rate=0.1, n_estimators=160, silent=True)
model.fit(X_train, y_train)

# 对测试集进行预测
ans = model.predict(X_test)

model.score(X_test,y_test)

0.9124320518930263

plt.figure()
gbm.plot_importance(model)
plt.show()

<Figure size 640x480 with 0 Axes>

plt.figure()
gbm.plot_tree(model)
plt.show()

<Figure size 640x480 with 0 Axes>

---