First tried with all features, then only light curve features, then with only first 2 light curve features in constants.py. Tried changing feature set within load_reference() and/or anomaly_detection(). Also happens with pca turned on in load_reference(). Some example objects:
- ZTF21aaublej
- ZTF21abbzjeq
- ZTF23aakrkjt
- ZTF23aapksof
All are standard SNe Ia marked with high > 60% anomaly scores, some reaching 99% almost immediately.
First tried with all features, then only light curve features, then with only first 2 light curve features in constants.py. Tried changing feature set within load_reference() and/or anomaly_detection(). Also happens with pca turned on in load_reference(). Some example objects:
All are standard SNe Ia marked with high > 60% anomaly scores, some reaching 99% almost immediately.