C16 D1: Derived Out-of-Fold Predictions and Output-Level Reproducibility Dataset

This record contains 1,152 out-of-fold predictions from six estimators on 192 DJI Matrice 100 flights across 17 collection dates; a 209-record label and eligibility ledger; date-grouped five-fold assignments; archived model-score and paired-date summaries; and provenance and data dictionary files. The recorded-window discharge-energy target is an operational voltage-current integral rather than an independently calibrated full-flight energy measurement. The release permits independent output-level arithmetic verification but does not include the original CMU telemetry, rerun model training, certify battery instrumentation, or establish external-platform generalization. Data from the original Carnegie Mellon University KiltHub dataset are attributed in ATTRIBUTION.md. The derived data are licensed under Creative Commons Attribution 4.0 International (CC BY 4.0).