Out-of-fold LightGBM recovers up to R²=0.982 of bias-corrected Hubble residual variance in LSST SN Ia mocks and R²=0.725 on DES 5YR, with consistent SHAP rankings, while redshift bins explain <1%.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
astro-ph.CO 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Machine Learning Closure Audits for LSST Photometric Supernova Cosmology
Out-of-fold LightGBM recovers up to R²=0.982 of bias-corrected Hubble residual variance in LSST SN Ia mocks and R²=0.725 on DES 5YR, with consistent SHAP rankings, while redshift bins explain <1%.