$\texttt{tdescore}$: An Accurate Photometric Classifier for Tidal Disruption Events

TL;DR

tdescore is a highly accurate photometric classifier for Tidal Disruption Events, capable of identifying TDEs from optical survey data with performance close to spectroscopic methods, and includes explainability features.

Contribution

We introduce tdescore, a novel binary photometric classifier for TDEs trained on ZTF data, with high accuracy and interpretability using SHAP explanations.

Findings

Rejects non-TDEs with 99.6% accuracy

Recalls 77.5% of TDEs with 80.2% precision

Outperforms existing photometric classifiers in the literature

Abstract

Optical surveys have become increasingly adept at identifying candidate Tidal Disruption Events (TDEs) in large numbers, but classifying these generally requires extensive spectroscopic resources. Here we present $\texttt{tdescore}$, a simple binary photometric classifier that is trained using a systematic census of $\sim$3000 nuclear transients from the Zwicky Transient Facility (ZTF). The sample is highly imbalanced, with TDEs representing $\sim$2% of the total. $\texttt{tdescore}$ is nonetheless able to reject non-TDEs with 99.6% accuracy, yielding a sample of probable TDEs with recall of 77.5% for a precision of 80.2%. $\texttt{tdescore}$ is thus substantially better than any available TDE photometric classifier scheme in the literature, with performance not far from spectroscopy as a method for classifying ZTF nuclear transients, despite relying solely on ZTF data and multi-wavelength catalogue cross-matching. In a novel extension, we use `SHapley Additive exPlanations' ($\texttt{SHAP}$) to provide a human-readable justification for each individual $\texttt{tdescore}$ classification, enabling users to understand and form opinions about the underlying classifier reasoning. $\texttt{tdescore}$ can serve as a model for photometric identification of TDEs with time-domain surveys, such as the upcoming Rubin observatory.