EVQ

Evolving Vector Quantization for Classification of On-Line Data Streams

Note

Inspired by this algorithm and several others, I created a new one called SEVQ: https://github.com/sylwekczmil/sevq

Inspiration

Implementation done by Sylwester Czmil based on pseudocode and algorithm description from:

E. Lughofer, "Evolving Vector Quantization for Classification of On-Line Data Streams," 2008 International Conference on Computational Intelligence for Modelling Control & Automation, 2008, pp.779-784, doi: 10.1109/CIMCA.2008.47.

Installation

# create venv and activate
# install algorithm
pip install evq

Example usage

Training and prediction one sample at a time

from evq.algorithm import EVQ

c = EVQ(number_of_classes=2, vigilance=0.1)
c.partial_fit([-2, -2], 1)
c.partial_fit([-1, -1], 0)
c.partial_fit([1, 1], 0)
c.partial_fit([2, 2], 1)

print(c.predict([0, 0]))  # 0
print(c.predict([3, 3]))  # 1
print(c.predict([-3, -3]))  # 1

Training and prediction on multiple samples

from evq.algorithm import EVQ

c = EVQ(number_of_classes=2, vigilance=0.1)
c.fit(
    [[-2, -2], [-1, -1], [1, 1], [2, 2]],
    [1, 0, 0, 1],
    epochs=1, permute=False
)

print(c.predict([[0, 0], [3, 3], [-3, -3]]))  # [0, 1, 1]