New Model Check List

Please follow the following check-list for committing new models to AutoMPC.

1. Verify Licensing Status Make sure that any code you are checking is publicly release-able and compatible with the BSD 3-clause license.

2. Update Documentation Create a docstring for your model which describes the algorithm, parameters and hyperparameters. Refer to existing models for a formatting example. Then, update docs/source/sysid.rst to include your new model. Finally, update the list in the README to include your model.

3. Add Unit Tests Update test/test_models.py to add a unit test for your model, following the examples in the file. The unit tests can cover one or more hyperparameter configs. Ensure that the choice of configs in the unit test covers all major code paths in the model. Once the unit test is added, generate the pre-computed model vectors, by adding your model to the list in the if __name__ == "__main__" block, and then running python test_models.py precompute <model_name>. This will generate precomputed model vectors in the precomputed/ directory. These should be checked into version control.

4. Run Benchmarks Run the modelling benchmark for benchmark and add to the results to the Modelling Benchmark wiki page.

5. Submit a Pull Request!