# CarbonQSAR

Supporting code for:

**"The Homodesmotic Reference: A Two-Regime Geometric Model for Cycloalkane Ring Strain"**

## Overview

This package provides Python scripts that reproduce the full-subset regression, LOOCV, and cut-off scan analyses reported in the manuscript. All thermochemical data (composite `G_CC` and `H_CC` values) are hard-coded in each script and derived from the SI tables.

## Files

| File                   | Description |
| `cutoff_scan.py` | Cut-off scan for the homodesmotic reaction (t = 5–14). |
| `loocv_analysis.py` | Leave-one-out cross-validation for two- vs. three-descriptor models. |
| `hydrogenation_qsar.py` | QSAR analysis of the hydrogenation ring-opening reaction (`cycloalkane + H2 -> n-alkane`). |

## Requirements

- Python 3.8+
- numpy
- scikit-learn

Install dependencies:

```bash
pip install numpy scikit-learn
```

For users in China, use the Tsinghua mirror for faster download:

```bash
pip install numpy scikit-learn -i https://pypi.tuna.tsinghua.edu.cn/simple
```

## Usage

Run any script directly:

```bash
python cutoff_scan.py
python loocv_analysis.py
python hydrogenation_qsar.py
```

## Output

Each script prints formatted tables to the console. No external input files are required.

## License

Provided for reproducibility and review purposes.