Use the requirements file to install all the necessary packages for using the scripts.
The som package contains the code for the Self-organizing map mapper. Use the two examples in the examples package for:
- classifying a set of random colors on a map. The Kohonen matrix is displayed with the colors of the nodes after training and the U-Matrix is used for displaying the mean distance to neighbors for each node.
- classifying a set of documents extracted from the 20-news-groups dataset. For comparison a similar classification is performed with inspiration on the scikit-learn example (http://scikit-learn.org/stable/auto_examples/text/document_classification_20newsgroups.html#example-text-document-classification-20newsgroups-py)
- Increase unit-tests coverage
- Improve profiling and accordingly cythonize functions that are bottlenecks
- Implement hexagonal maps training
- Improve the interface to classification metrics (topological error, quantization error...)
- Kohonen Publication : http://scholar.google.fr/scholar_url?url=http://users.ics.aalto.fi/wsom97/progabstracts/ps/kohonen_2.ps&hl=fr&sa=X&scisig=AAGBfm2hcJeT6U5-RY3Yp8XrF3JqUupGEQ&nossl=1&oi=scholarr&ved=0CCAQgAMoADAAahUKEwjQof34nPLIAhUJfhoKHYlzB4k
- PyMVPA implementation : http://www.pymvpa.org/examples/som.html