SCDR delivers asymptotically guaranteed conditional coverage for time-series predictions by conformalizing estimated highest-density regions with a quantile random forest adjustment that adapts to non-exchangeability.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
stat.ME 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Conformal Prediction with Time-Series Data via Sequential Conformalized Density Regions
SCDR delivers asymptotically guaranteed conditional coverage for time-series predictions by conformalizing estimated highest-density regions with a quantile random forest adjustment that adapts to non-exchangeability.