README.md

Kohonen-Self-Organizing-Maps-SOM

Implementation of Kohonen Self-Organizing Maps (SOM) for two dimensions

Table of Contents

1. About the project
2. Built With
3. Generated Maps
4. License

About The Project

This task was made for course "Data Analysis". We were given some sample data (here represented by circles) and we had to implement Kohenen's SOM in order to achieve basic clustering. In program we can choose different parameters to fit models that vary in data lenght. It is also possible to change circles (data) number and location.

Below I presented some of the maps, that can be generated using this code:

(back to top)

Some generated maps:

• 18 neurons - fitting parameters (radius: 0.15, learning rate: 0.15)

• 18 neruons - radius is too large (radius: 0.5, learning rate: 0.15)

• 8 neurons - correct parameters (radius: 0.15, learning rate: 0.01)

• 8 neurons - learning rate is too small (radius: 0.15, learning rate: 0.001)

Built With

• Python
• Pandas
• Numpy
• Matplotlib

(back to top)

License

Distributed under the MIT License. See LICENSE.txt for more information.

(back to top)