Pattern-based Embedding for Time Series Classification (PETSC).
This repository contains a Python/C++ implementation of PETSC, that is compatible with aeon. Two variants of PETSC are supported as well: MR-PETSC and PETSC-SOFT.
The easiest way to install petsc-ts is via pip:
pip install petsc-ts
If you want to run the tests locally (currently only an aeon test to check if the estimator complies with the required format), install the optional dependency:
pip install "petsc-ts[test]"
You can then import the PetsClassifier class as follows and use it as any other scikit-learn-compatible estimator:
from petsc_ts.classifier import PetsClassifier
clsf = PetsClassifier()
clsf.fit(X_train, y_train)
clsf.predict(X_test)
clsf.predict_proba(X_test)