K-Means

This repository was prepared as an educational project for lectures of Artificial Intelligence at Smíchovská střední průmyslová škola a gymnázium.

K-Means is a popular cluster algorithm widely used in various fields for data analysis. Its centroid-based nature enables the simple way of assignment of each given instance into any of the found clusters.

To find more information about this algorithm, have a look at the Wikipedia page.

This implementation is not meant to be used in production environments, because it uses just the simple Python means; no usage of NumPy or any other scientific and compute performance optimized library. The purpose of this project is to be rather descriptive and not for being used in any mission-critical applications.

Requirements

The only dependency required by this project is a pytest library for testing purposes. If you are not about to run these tests, you can skip it. Otherwise, clone this repository to your device (into a virtual environment, if you want) and run:

  python -m pip install --upgrade pip
  python -m pip install pytest

On some devices, this might be kinda problematic - for example on linux OS. So if the previous does not work, try these commands:

  python3 -m pip install --upgrade pip
  python3 -m pip install pytest

Project structure

This project is organized into simple multi-directory structure as seen below:

• .github - contains just a simple workflow to run all the tests using GitHub Actions

• src - contains the actual source code, the implementation

• test - contains the unit tests to validate some of the features

Notable objects

The most important classes are definitely:

• Point and Centroid (both defined in the ./src/datapoint.py module),

• abstract class Metric used to calculate the distance between two points in a multidimensional space (./src/datapoint.py)

• class KMeans, that defines instances providing the actual model (this is defined in the ./src/k_means.py module).