Explainability Insights
Review the biggest churn drivers using SHAP and LIME with the active data source and model context in view.
Built-in modelPre-trained XGBoost model
Insight Context
Default datasetModel confirmed
Dataset
IBM Telco churn dataset
Model
Pre-trained XGBoost model
Insight interpretation follows the current workspace selection so users can see which dataset and model the page is anchored to.
SHAP Feature Importance
Positive values push predictions toward churn. Negative values pull them toward retention.
How to read SHAP
Positive SHAP means a feature is increasing churn risk.
Negative SHAP means a feature is helping retention.
The larger the absolute value, the stronger the influence on the model decision.
Top Churn Risks
Contract_Month-to-month+0.380
MonthlyCharges+0.220
InternetService_Fiber+0.180
TechSupport_No+0.140
OnlineSecurity_No+0.110