Code for the paper called ``AuToMATo: An Out-Of-The-Box Persistence-Based Clustering Algorithm''.
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 pyproject.toml. Developer dependencies indicate those that are needed only in order to run the scripts that reproduce results and figures from the paper, as opposed to using AuToMATo on its own. Provided that uv is installed, these dependencies can be installed by running uv pip instal -r pyproject.toml (to exclude dev-dependencies) and uv pip install --group dev -r pyproject.toml (to include dev-dependencies). The environment specified in uv.lock can be recreated by running uv sync --no-dev and uv sync, respectively.
Example of installing AuToMATo for uv users
$ git clone url/to/repo
$ cd AuToMATo
$ uv sync --no-dev
$ source .venv/bin/activate
$ python
>>> from automato import Automato
>>> ...