How ranuth-nanvidu/cancer-ml-leakage-demo works
This is a reproducible cancer-classification data-leakage experiment, not a deployed prediction service. Users explore WDBC, compare duplicate-before-split and split-before-duplicate experiments, then inspect mean, maximum, and minimum accuracy results. The notebooks are the documented user workflow; the importable package supplies data loading, split strategies, classifiers, and experiment metrics. Edges follow README descriptions or shown source calls; notebook internals were not sampled, so their connections reflect documented behavior only.
Main entry points: 02_leaky_pipeline.ipynb Leaky experiment · 03_correct_pipeline.ipynb Correct experiment · results Tables and figures · 04_max_vs_min_accuracy.ipynb Accuracy comparison
Components 10
Notebook workflow
- Dataset exploration
notebooks/01_eda.ipynb - Leaky experiment
notebooks/02_leaky_pipeline.ipynb - Correct experiment
notebooks/03_correct_pipeline.ipynb - Accuracy comparison
notebooks/04_max_vs_min_accuracy.ipynb
Experiment logic
- WDBC loading and split
src/cancer_ml_leakage_demo/data.py - Duplicate split strategies
src/cancer_ml_leakage_demo/leakage.py - Repeated evaluation
src/cancer_ml_leakage_demo/metrics.py - Seven classifiers
src/cancer_ml_leakage_demo/models.py
Analysis outputs
- Tables and figures
results
Other components
- Researcher
Connections 16
- Researcher → Dataset exploration: explores data
README.md - Researcher → Leaky experiment: runs experiment
README.md - Researcher → Correct experiment: runs experiment
README.md - Researcher → Accuracy comparison: reviews comparison
README.md - Leaky experiment → WDBC loading and split: loads data
README.md - Correct experiment → WDBC loading and split: loads data
README.md - Leaky experiment → Duplicate split strategies: uses leaky split
README.md - Correct experiment → Duplicate split strategies: uses corrected split
README.md - Leaky experiment → Repeated evaluation: runs evaluation
README.md - Correct experiment → Repeated evaluation: runs evaluation
README.md - Repeated evaluation → Seven classifiers: creates models
src/cancer_ml_leakage_demo/metrics.py - Duplicate split strategies → WDBC loading and split: uses simple split
src/cancer_ml_leakage_demo/leakage.py - Leaky experiment → Tables and figures: writes summary
README.md - Correct experiment → Tables and figures: writes summary
README.md - Accuracy comparison → Tables and figures: reads summaries
README.md - Accuracy comparison → Tables and figures: creates comparison
README.md