有没有办法在 SHAP 蜂群图中自定义特征顺序？

Question

我想知道是否有办法更改 SHAP 蜂群图中特征的显示顺序。文档描述了“转换”，例如使用 shap_values.abs 或 shap_values.abs.mean(0) 来更改方式排序是计算出来的，但我真正想要的是放入一个特征或索引列表并按此排序。

来自文档：

shap.plots.beeswarm(shap_values, order=shap_values.abs)

This is the resulting plot

Answer 1

这是排序的默认实现：

import xgboost
import shap

X, y = shap.datasets.adult()
model = xgboost.XGBClassifier().fit(X, y)

explainer = shap.Explainer(model, X)
shap_values = explainer(X)

shap.plots.beeswarm(shap_values, max_display=12, order=shap.Explanation.abs.mean(0))

然后，如果您想手动定义输出列的顺序：

order = [
    "Country",
    "Workclass",
    "Education-Num",
    "Marital Status",
    "Occupation",
    "Relationship",
    "Race",
    "Sex",
    "Capital Gain",
    "Capital Loss",
    "Hours per week",
    "Age",
]
col2num = {col: i for i, col in enumerate(X.columns)}

order = list(map(col2num.get, order))

shap.plots.beeswarm(shap_values, max_display=12, show=False, color_bar=False, order=order)
plt.colorbar()
plt.show()

有没有办法在 SHAP 蜂群图中自定义特征顺序？

Is there a way to customize the feature order in a SHAP beeswarm plot?

python

plot

matplotlib

beeswarm

shap