Benchmarks
==========
    
To generate the data to run all the benchmarks: ``pixi run benchmark-generate-matrices``. Then, to run all the benchmarks with default settings: ``python src/tabmat/benchmark/benchmark-run``. To produce or update these figures, open ``src/tabmat/benchmark/visualize_benchmarks.py`` as a notebook via ``jupytext``. 

For more info on the benchmark CLI: ``pixi run python src/tabmat/benchmark/main.py --help``.

Performance
^^^^^^^^^^^

Dense matrix, 4M x 10:

.. image:: _static/dense_bench.png
   :width: 700

One-hot encoded categorical variable, 1M x 100k:

.. image:: _static/one_cat_bench.png
   :width: 700

Sparse matrix, 1M x 1k:

.. image:: _static/sparse_bench.png
   :width: 700

Two categorical matrices, 1M x 2k:

.. image:: _static/two_cat_bench.png
   :width: 700

Dense matrix plus two categorical matrices, 3M x (dense=5, cat1=10, cat2=1000).

.. image:: _static/dense_cat_bench.png
   :width: 700