Code for the paper called ``AuToMATo: A Parameter-Free Persistence-Based Clustering Algorithm'' by Huber, Kališnik and Schnider.
To run the scripts reproducing the results and figures in the paper, run python3 -m eval.eval, python3 -m mapper_applications.make_diabetes_pictures or python3 -m mapper_applications.make_synth_pictures.
If the TTK-clustering algorithm is to be included in evaluation (uncomment relevant lines in eval.py to do so), ParaView and the Topology ToolKit must be installed.
Example of running AuToMATo
>>> from automato import Automato
>>> from sklearn.datasets import make_blobs
>>> X, y = make_blobs(centers=2, random_state=42)
>>> aut = Automato(random_state=42).fit(X)
>>> aut.n_clusters_
2
>>> (aut.labels_ == y).all()
True
Requirements
Required Python dependencies are specified in Pipfile and in requirements.txt and requirements-dev.txt. Developer dependencies indicate those that are needed only when running the scripts that reproduce results and figures from the paper. Dependencies from Pipfile can be installed by running pipenv install or pipenv install --dev (assuming that Pipenv is installed on the system).
Example of installing AuToMATo for pipenv users
$ git clone https://github.com/m-a-huber/AuToMATo
$ cd AuToMATo
$ pipenv install
$ pipenv shell
$ python
>>> from automato import Automato
>>> ...