Cross-Check Cost/Quality Framework – Reproducibility Package

This repository accompanies the paper
“Framework for Optimizing Cross-Check Procedures” (ProMAC 2025).

It provides

• fully reproducible Python scripts (src/)
• a minimal data sample (data/sample_changes.json)
• a one-liner wrapper (run.sh) that sets up a virtual environment via python3 venv,
  runs the CLI script, and launches the updated Streamlit app with full parameterized sensitivity analysis.

Quick Start (one line)

The script will

• delete and recreate .venv/ (⚠ all contents will be lost)
• compute and display cross-check efficiency metrics per project
• launch an interactive Streamlit dashboard with:
  • Elasticity and Standardized Sensitivity analysis
  • Tornado diagrams
  • Monte Carlo distributions
  • Configurable display language (EN/JP)
  • Unified layout and updated parameter names

Tested on macOS 14 (Python 3.9) and Ubuntu 22.04 (Python 3.10).

Command-Line Arguments (CLI)

The CLI version accepts the following optional arguments:

• streamlit — Launch only the Streamlit app
• gerrit — Run Gerrit-related CLI extraction only
• all (default) — Run both CLI and Streamlit in one go

Examples:

./run.sh streamlit
./run.sh gerrit

Streamlit App Overview

The Streamlit dashboard displays:

• Efficiency Metrics (E_total, C, S) — per project or scenario
• Sensitivity Analysis
  • Relative Sensitivity (∂E/∂x × x/E)
  • Standardized Sensitivity (∂E/∂x × σₓ / σ_E)
• Tornado Diagrams (±20% variation)
• Monte Carlo Simulation with histogram and spider charts
• Language Toggle (EN/JA) — upper-right language switch
• Scenario Filters — quality/schedule grid (2×2), with visual comparison

All graphs are consistently styled and labeled, with markdown hints in both languages.

Gerrit Integration

The CLI version automatically reads recent Gerrit logs (e.g. data/sample_changes.json) and:

• extracts submission and review data
• computes work vs review durations
• exports summary metrics for visualization
• serves as reproducible backend for the GUI

To update the dataset, replace data/sample_changes.json with your own JSON export from Gerrit.

Notes

• If you later re-enable pyenv, simply replace the first block with the previous pyenv logic; everything else stays the same.

• The Streamlit app reads Gerrit live each time, so the GUI reflects the same data that the CLI just printed.

• The latest version reflects improvements in visualization clarity, parameter control,
  and reproducibility for academic use.