MaatPy

The MaatPy toolbox is written using Python programming language and is deployed as a package. The MaatPy package collects several algorithms that together support an analysis pipeline for imbalanced datasets. It includes novel implementations for:

• SMOTEBoost
• SMOTEBagging
• AdaBoost cost sensitive variants AdaCost, AdaC1, AdaC2 and AdaC3
• Balanced Random Forest Classifier.

as well as modifications for the imblearn combination samplers:

• SMOTEENN
• SMOTETomek

Installation

---

Dependencies

scikit-learn requires:

• imbalanced-learn>=0.3.3
• joblib>=0.12.2
• matplotlib>=2.2.2
• numpy>=1.15.0
• pandas>=0.23.4
• scikit-learn>=0.19.2
• scipy>=1.1.0

User installation

The easiest way to install this package is clone this repository:

git clone https://github.com/gkapatai/MaatPy/new/master?readme=1

and then run:

python setup.py install

This will install the package and all it's dependencies.

Testing

After installation you can launch the test suite by running the following command from within the root directory of this repository:

nosetests tests

This requires prior installation of the nose python module:

pip install nose