Metadata-Version: 2.1
Name: cubist
Version: 0.0.10
Summary: A Python port of the R Cubist library.
Home-page: https://github.com/pjaselin/Cubist
Author: Patrick Aselin
License: LICENSE
Platform: UNKNOWN
Classifier: Programming Language :: Python
Classifier: License :: OSI Approved :: GNU General Public License v3 (GPLv3)
Classifier: Topic :: Software Development :: Libraries
Classifier: Intended Audience :: Developers
Classifier: Intended Audience :: Science/Research
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: numpy (>=1.19.2)
Requires-Dist: pandas (>=1.1.3)
Requires-Dist: scikit-learn (>=0.24.2)

# Cubist

A Python package for fitting Ross Quinlan's [Cubist](https://www.rulequest.com/cubist-unix.html) v2.07 regression model. Inspired by and based on the [R wrapper](https://github.com/topepo/Cubist) for Cubist. Designed after and inherits from the [scikit-learn](https://scikit-learn.org/stable/) framework.

## Background
Cubist is a novel regression algorithm develped by Ross Quinlan...

## Use
```python
>>> from cubist import Cubist
>>> model = Cubist()
>>> model.fit(X, y)
>>> model.predict(X)
```

## Benchmarks
From literature, there examples of Cubist outperforming RandomForest and other boosted models, to demonstrate this, the following benchmarks are provided to compare models. The scripts that achieved these results are provided in the benchmarks folder.

## Building
```bash
python -m build --sdist --wheel .
```

## Installing from Source
```bash
pip install --upgrade .
```

## Interesting Links:  
- https://www.linkedin.com/pulse/machine-learning-example-r-using-cubist-kirk-mettler
http://rulequest.com/cubist-info.html

## Literature
- https://sci2s.ugr.es/keel/pdf/algorithm/congreso/1992-Quinlan-AI.pdf
- http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.34.6358&rep=rep1&type=pdf

