ELAsTiCC 1.0.0

• Compute features for training set and pipeline
• Model

Release 1.2.3-P

• Renamed the model to hierarchical_rf_1.1.0

Release 2.0.3

Fix model pkl files

Fix installation

• Updated requirements
• Updated setup

Now you need to install Cython, numpy and turbofats, mhps and P4J before installing this package.

Fixed compute of probabilities

Fixed order of operations in compute of probabilities (models.py):

        prob_children = []
        resp_children = {}
        child_models = [
            self.stochastic_classifier,
            self.transient_classifier,
            self.periodic_classifier,
        ]
        child_names = ["Transient", "Stochastic", "Periodic"]

        for name, model in zip(child_names, child_models):
            prob_child = pd.DataFrame(
                model.predict_proba(input_features),
                columns=model.classes_,
                index=input_features.index,
            )

            resp_children[name] = prob_child
            prob_child = prob_child.mul(prob_root[name].values, axis=0)
            prob_children.append(prob_child)
        prob_all = pd.concat(prob_children, axis=1, sort=False)

Update classifier model 1.1.0

The tag with the suffix -P means this is designed for the processing pipeline.

Changelog:

• Update classifier model to 1.1.0 (July 2021)
• Removed StreamDetectionsPreprocessorZTF
  • CustomStreamHierarchicalExtractor uses DetectionsPreprocessorZTF
  • StreamedForcedPhotometryExtractor uses DetectionsPreprocessorZTF

Fix Preprocessing For Pipeline

This version with suffix -P means it is specifically directed for the stream Pipeline.

Changelog:

• Fixes preprocessing
  (n_det_g + n_det_r) >= 6 changed to (n_det_g >= 6) || (n_det_g >= 6)

New interface

Merge pull request #11 from alercebroker/library_refactor

Library refactor: v2.0.0

Release 1.1.1

Minor fix that lowers the requirements on numpy down to 1.16.0

Release 1.1.0

Release 1.1.0

• Uses batches