PL-kNN: A Python-based implementation of a parameterless k-Nearest Neighbors classifier [Formula presented]
Loading...
Files
External sources
External sources
Date
Advisor
Coadvisor
Graduate program
Undergraduate course
Journal Title
Journal ISSN
Volume Title
Publisher
Type
Article
Access right
Files
External sources
External sources
Abstract
This paper presents an open-source implementation of PL-kNN, a parameterless version of the k-Nearest Neighbors algorithm. The proposed model, developed in Python 3.6, was designed to avoid the choice of the k parameter required by the standard k-Nearest Neighbors technique. Essentially, the model computes the number of nearest neighbors of a target sample using the data distribution of the training set. The source code provides functions resembling the Scikit-learn methods for fitting the model and predicting the classes of the new samples. The source code is available in the GitHub repository with instructions for installation and examples for usage.
Description
Keywords
Classification, Clustering, k-Nearest Neighbors, Machine learning, Python
Language
English
Citation
Software Impacts, v. 15.
