oximachinerunner

oximachine for AiiDA lab: Core functionalities of oximachine with stripped dependencies.

Warning: This model works excellent on a test set but it might give fully unphysical predictions in some cases. Consider it in alpha phase

• It is good to know where it fails

• We work on improving the model by training it on a larger subset of the CSD with a new architecture

• The featurization can be slow in some cases. In practice, it is best to get the smallest possible cell of a clean structure.

• There is still one dependency on one of my forks of a well-known package.

• The package is slow

• For compatability and reproducibility we need to pin an old scikit-learn version

features not deployed for AiIDA lab

• Feature importance (slow as it has to integrate the dataset. Also, it is quite likely that we will break the API here in the future when we add new features)
• Most similar structures in training set (is typically fast though, as it uses a KDtree)
• Uncertainity estimate (not sure how the best way to use this in a workchain?)

assets

• votingclassifier.joblib is the model that is currently deployed. It is a voting classifier with four different base estimators
• scaler.joblib is the standard scaler
• KAJZIH_freeONLY.cif and ACODAA.cif are some test structures.