如何访问 scikit-learn 管道中的“每个估计器”？ [关闭]

Question

如何在管道中访问"Log"？

pipelines = {
    "Log": Pipeline(
        [("scl", StandardScaler()), ("est", LogisticRegression(random_state=1))]
    ),
    "Rf": Pipeline([("est", RandomForestClassifier(random_state=1))]),
    "Rf_Pipeline": Pipeline(
        [
            ("scl", StandardScaler()),
            ("reduct", PCA(n_components=10, random_state=1)),
            ("est", RandomForestClassifier(random_state=1)),
        ]
    ),
}


Pipelines.item(Log)

目前我得到：

NameError: name 'Log' is not defined

Answer 1

管道对象可以被视为字典。在您的情况下，您已将多个管道存储到字典中。要访问不同的密钥（管道），您只需使用 dict['key'] 或 dict.get['key']。

1. 对于第一级（子管道），只需使用dict['key']
2. 对于第二级（子管道中的步骤），您可以再次使用named_steps 获取带有步骤的dict，然后以相同的方式引用每个步骤。

这是代码-

from sklearn.pipeline import Pipeline
from sklearn.preprocessing import StandardScaler
from sklearn.linear_model import LogisticRegression
from sklearn.decomposition import PCA
from sklearn.ensemble import RandomForestClassifier

pipelines = {
    "Log": Pipeline(
        [("scl", StandardScaler()), ("est", LogisticRegression(random_state=1))]
    ),
    "Rf": Pipeline([("est", RandomForestClassifier(random_state=1))]),
    "Rf_Pipeline": Pipeline(
        [
            ("scl", StandardScaler()),
            ("reduct", PCA(n_components=10, random_state=1)),
            ("est", RandomForestClassifier(random_state=1)),
        ]
    ),
}

first_subpipeline = pipelines['Log']
second_step_first_subpipeline =  pipelines['Log'].named_steps['est']

print(first_subpipeline)
print(second_step_first_subpipeline)

Pipeline(steps=[('scl', StandardScaler()),
                ('est', LogisticRegression(random_state=1))])

LogisticRegression(random_state=1)