Stability-Aware Adaptive Focal Evidence Clustering: Synthetic Controlled Benchmark, Public-Dataset Workflow and Reference Implementation
Description
This repository accompanies a study of stability-aware adaptive focal evidence clustering. It contains a deterministic, 516-observation synthetic benchmark representing cluster cores, pair bridges, a junction and outliers; reference Python code for the clustering and comparison methods; scripts that access the Iris, Wine and Wisconsin Diagnostic Breast Cancer datasets through scikit-learn; and aggregate outputs with validation checks. The public datasets are accessed through their loaders and are not redistributed.
Files
Steps to reproduce
1.Download and extract the archive. Python 3.12 is recommended. 2.In the extracted archive’s top-level directory, run: python -m pip install -r requirements.txt 3.Run: python run_reproduction.py 4.Run: python validate_outputs.py 5.Inspect data/generated/controlled_benchmark.csv and outputs/validation_report.json. The scripts regenerate the controlled benchmark and public/controlled experiment outputs; they do not reproduce the confidential retail analysis.
Institutions
- Istanbul Health and Technology UniversityIstanbul, Istanbul