Embedded Files
Code, Results, and Reproducibility
All results are reproducible. Set random_state=42 for identical outputs.
Key Implementation Snippets
Key Implementation Snippets
Execution of the SMOTE-resampled training pipeline within the Google Colab environment. This view demonstrates the integrated workflow, showing the transition from data preprocessing to the generation of the EU AI Act-compliant Transparency Record.
A global explanation of the Random Forest model using SHAP (SHapley Additive exPlanations). This plot visualizes how features like V14 and V17 impact the model’s fraud prediction across the entire dataset, ranking features by their total contribution to the output.
Local Interpretable Model-agnostic Explanations (LIME) for a single transaction. This chart provides 'case-level' transparency by showing which specific feature values pushed the model toward a 'Fraud' or 'Legitimate' classification for an individual record.
Page updated
Report abuse