rank-similarity 0.0.1

Rank Similarity is a set of non-linear classification and transform tools for large datasets.

Rank Similarity

Rank Similarity is a set of non-linear classification and transform tools for large multi-dimensional datasets that use the scikit-learn API.

Installation

Dependencies

rank-similarity requires:

• Scikit-learn (>= 0.23)
• Python (>= 3.7)
• NumPy (>= 1.14.6)
• SciPy (>= 1.1.0)

Optionally for plotting examples:

• matplotlib (>= 2.2.2)

Install via pip

pip install rank-similarity

Development version

To get the latest development version, clone the GitHub repository:

git clone https://github.com/KatharineShapcott/rank-similarity

Usage

from ranksim import RankSimilarityClassifier
X = [[0, 1], [1, 0]]
y = [0, 1]
clf = RankSimilarityClassifier()
clf.fit(X, y)
pred = clf.predict(X)

More Information

Documentation

More details and background information is available in the online documentation.

License

The package is new BSD licensed.

Citation

Please cite the following publication (in preparation) [1].

[1] Shapcott, Bird, & Singer. Confusion-based rank similarity filters for computationally-efficient machine learning on high dimensional data. In preperation. (2021)