hero

SuSi – Self-organizing Maps with Python

Self-organizing maps (SOM) Python package for unsupervised, supervised and semi-supervised learning

Get Started → (opens new window)

Simple syntax

You can use SuSi similar to the well-known scikit-learn syntax, e.g., fit(), predict(), transform().

Multiple applications

Support for unsupervised clustering and visualization, semi-supervised and supervised classification and regression.

Built-in visualization

Present your data with SuSi through built-in plotting scripts and example notebooks.

# What is SuSi?

SuSi is a Python package for unsupervised, supervised, and semi-supervised learning. It is built as an estimator in scikit-learn (opens new window) style and works with all currently-maintained Python 3 (opens new window) versions.

This is a basic example on how to use SuSi for supervised classification:

import susi

# load your dataset
X_train, X_test, y_train, y_test = ...

# initialize and fit SuSi
som = susi.SOMClassifier()
som.fit(X_train, y_train)

# predict and calculate the accuracy score
y_pred = som.predict(X_test)
print(som.score(X_test, y_test))

# Getting started

Installation of SuSi via pip:

pip install susi

Installation of SuSi via conda-forge:

conda install -c conda-forge susi