调优随机森林的超参数

Python 树模型机器学习

Elie Kawerk

Data Scientist

随机森林的超参数

CART 超参数
基学习器数量
自助采样（bootstrap）
……

调优成本高

超参数调优：

计算开销大，
有时仅带来轻微提升，

权衡调优对整个项目的影响。

查看 sklearn 中的 RF 超参数

# 导入 RandomForestRegressor 
from sklearn.ensemble import RandomForestRegressor

# 设定随机种子，保证可复现
SEED = 1

# 实例化随机森林回归器 'rf' 
rf = RandomForestRegressor(random_state= SEED)

# 查看 rf 的超参数
rf.get_params()

{'bootstrap': True,
 'criterion': 'mse',
 'max_depth': None,
 'max_features': 'auto',
 'max_leaf_nodes': None,
 'min_impurity_decrease': 0.0,
 'min_impurity_split': None,
 'min_samples_leaf': 1,
 'min_samples_split': 2,
 'min_weight_fraction_leaf': 0.0,
 'n_estimators': 10,
 'n_jobs': -1,
 'oob_score': False,
 'random_state': 1,
 'verbose': 0,
 'warm_start': False}

# 基本导入
from sklearn.metrics import mean_squared_error as MSE
from sklearn.model_selection import GridSearchCV

# 定义超参数网格 'params_rf'
params_rf = {
              'n_estimators': [300, 400, 500],
              'max_depth': [4, 6, 8],
              'min_samples_leaf': [0.1, 0.2],
              'max_features': ['log2', 'sqrt']
             }

# 实例化 'grid_rf'
grid_rf = GridSearchCV(estimator=rf,
                       param_grid=params_rf, 
                       cv=3,
                       scoring='neg_mean_squared_error',
                       verbose=1,
                       n_jobs=-1)

搜索最佳超参数

# 在训练集上拟合 'grid_rf'
grid_rf.fit(X_train, y_train)

对每个 36 个候选进行 3 折拟合，共 108 次拟合
[Parallel(n_jobs=-1)]: Done  42 tasks      | elapsed:   10.0s
[Parallel(n_jobs=-1)]: Done 108 out of 108 | elapsed:   24.3s finished
RandomForestRegressor(bootstrap=True, criterion='mse', max_depth=4,
           max_features='log2', max_leaf_nodes=None,
           min_impurity_decrease=0.0, min_impurity_split=None,
           min_samples_leaf=0.1, min_samples_split=2,
           min_weight_fraction_leaf=0.0, n_estimators=400, n_jobs=1,
           oob_score=False, random_state=1, verbose=0, warm_start=False)

提取最佳超参数

# 从 'grid_rf' 中提取最佳超参数
best_hyperparams = grid_rf.best_params_

print('最佳超参数:\n', best_hyperparams)

最佳超参数:
        {'max_depth': 4,
         'max_features': 'log2', 
         'min_samples_leaf': 0.1,
         'n_estimators': 400}

评估最佳模型性能

# 从 'grid_rf' 中提取最佳模型
best_model = grid_rf.best_estimator_
# 预测测试集标签
y_pred = best_model.predict(X_test)
# 评估测试集 RMSE
rmse_test = MSE(y_test, y_pred)**(1/2)
# 打印测试集 RMSE
print('rf 测试集 RMSE: {:.2f}'.format(rmse_test))

rf 测试集 RMSE: 3.89

Passons à la pratique !

Python 树模型机器学习