可解释分类

在本 Jupyter Notebook 中，我们将拟合分类可解释增强机器 (EBM)、逻辑回归 (LogisticRegression) 和分类树 (ClassificationTree) 模型。拟合完成后，我们将利用它们 Glassbox 的特性来理解其全局和局部解释。

本 Notebook 位于 GitHub 上的我们的示例文件夹中。

# install interpret if not already installed
try:
    import interpret
except ModuleNotFoundError:
    !pip install --quiet interpret pandas scikit-learn

import numpy as np
import pandas as pd
from sklearn.model_selection import train_test_split
from interpret import show
from interpret.perf import ROC

from interpret import set_visualize_provider
from interpret.provider import InlineProvider
set_visualize_provider(InlineProvider())

df = pd.read_csv(
    "https://archive.ics.uci.edu/ml/machine-learning-databases/adult/adult.data",
    header=None)
df.columns = [
    "Age", "WorkClass", "fnlwgt", "Education", "EducationNum",
    "MaritalStatus", "Occupation", "Relationship", "Race", "Gender",
    "CapitalGain", "CapitalLoss", "HoursPerWeek", "NativeCountry", "Income"
]
X = df.iloc[:, :-1]
y = (df.iloc[:, -1] == " >50K").astype(int)

seed = 42
np.random.seed(seed)
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.20, random_state=seed)

探索数据集

from interpret.data import ClassHistogram

hist = ClassHistogram().explain_data(X_train, y_train, name='Train Data')
show(hist)

训练可解释增强机器 (EBM)

from interpret.glassbox import ExplainableBoostingClassifier

ebm = ExplainableBoostingClassifier()
ebm.fit(X_train, y_train)

ExplainableBoostingClassifier()

在 Jupyter 环境中，请重新运行此单元格以显示 HTML 表示或信任该 Notebook。
在 GitHub 上，HTML 表示无法渲染，请尝试使用 nbviewer.org 加载此页面。

EBM 是 Glassbox 模型，因此我们可以编辑它们

# post-process monotonize the Age feature
ebm.monotonize("Age", increasing=True)

ExplainableBoostingClassifier()

在 Jupyter 环境中，请重新运行此单元格以显示 HTML 表示或信任该 Notebook。
在 GitHub 上，HTML 表示无法渲染，请尝试使用 nbviewer.org 加载此页面。

全局解释：模型整体学到了什么

ebm_global = ebm.explain_global(name='EBM')
show(ebm_global)

局部解释：个体预测是如何做出的

ebm_local = ebm.explain_local(X_test[:5], y_test[:5], name='EBM')
show(ebm_local, 0)

评估 EBM 性能

ebm_perf = ROC(ebm).explain_perf(X_test, y_test, name='EBM')
show(ebm_perf)

让我们测试一些其他可解释模型

from interpret.glassbox import LogisticRegression, ClassificationTree

# We have to transform categorical variables to use Logistic Regression and Decision Tree
X = pd.get_dummies(X, prefix_sep='.').astype(float)
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.20, random_state=seed)

lr = LogisticRegression(random_state=seed, penalty='l1', solver='liblinear')
lr.fit(X_train, y_train)

tree = ClassificationTree()
tree.fit(X_train, y_train)

<interpret.glassbox._decisiontree.ClassificationTree at 0x7f0749bf3940>

使用 Dashboard 比较性能

lr_perf = ROC(lr).explain_perf(X_test, y_test, name='Logistic Regression')
show(lr_perf)

tree_perf = ROC(tree).explain_perf(X_test, y_test, name='Classification Tree')
show(tree_perf)

Glassbox：我们所有的模型都具有全局和局部解释

lr_global = lr.explain_global(name='Logistic Regression')
show(lr_global)

tree_global = tree.explain_global(name='Classification Tree')
show(tree_global)