Data Pipeline

Load, check, and prepare data

• Basic Pipeline: Load, Summary, Prepare
• Quality Check: Integrity, Outlier, Data drift
• Feature selection
• Exploratory data analysis

Interpretable Models

Inherent interpretability

• Main effect models: GLM, GAM, XGB1
• Interaction models: EBM, XGB2, GAMI-Net
• Local interpretable models: Tree, FIGS, ReLU-DNN

Post-hoc Explainability

Global and local explainability

• Global importance: PFI, H-statistic
• Global dependence: PDP, ALE
• Local methods: ICE, LIME, SHAP

Diagnostic Suite

Model validation and outcome testing

• Basic Tests: Accuracy, Weakspot, Overfit
• 3R Tests: Reliability, Robustness, Resilience
• Fairness test
• Segmented test
• Scored test

Model Comparison

Benchmarking through diagnostics

• Regression models
• Binary classification models
• Model fairness comparison

Low-Code Case Studies

PiML workflow and experimentation

• Example: Bikesharing Data
• Example: CaliforniaHousing Data
• Example: TaiwanCredit Data
• Fairness Simulation Study 1
• Fairness Simulation Study 2