panpapag / clustering-for-vectors-and-polygonal-curves

Implementation of 8 different variants of clustering algorithms for vectors and polygonal curves in C++

Clustering-for-vectors-and-polygonal-curves

This project implements clustering algorithms for in Euclidian space R^d and polygonal curves in Euclidian space R² using using the 8 combinations of the following variants. The Manhattan metric (L1) is used for vectors and the pseudo Dynamic Time Warping (DTW) for the curves.

Initialization

1. Random selection of K points/curves (simplest)
2. K-means++

Assignment

1. Lloyd's assignment (simplest approach)
2. Assignment by Range Search using LSH for vectors/curves (inverse assignment)

Update

1. Partitioning around medoids (PAM) `a la Lloyds
2. Calculating Mean Vector/DTW Centroid Curve

Installing

Download source code by typing:

https://github.com/PanPapag/Clustering-for-vectors-and-polygonal-curves.git

License

This project is licensed under the MIT License - see the LICENSE file for details