Reproducibility package for a Conditional, Boundary-Aware, and Auditable Deletion-Effect Weighting Framework for MCDM with Positive Decision Matrices and a Fixed Set of Alternatives

Abstract

This software archive contains the MATLAB source code, deterministic inputs, retained numerical records, manuscript-facing CSV outputs, figure-generation drivers, validation tests, and machine-readable provenance records for the associated study: “A Conditional, Boundary-Aware, and Auditable Deletion-Effect Weighting Framework for MCDM with Positive Decision Matrices and a Fixed Set of Alternatives.” The package implements Log-Ratio Composite Normalization (LRCN), LRCN-conditioned Deletion-Effect Weighting (LRCN-DEW), capped deletion-loss diagnostics, resolution-aware admissibility checks, candidate-set and reference-boundary audits, and the continue/disclose/stop reporting protocol. The archive includes the paired uncertainty records for Supplementary Table S6e, the external measurement-scale audit for Supplementary Table S7a, the independent boundary_provenance.csv record, manuscript figure assets, MATLAB tests, manuscript-consistency checks, and SHA-256 file checksums. The retained analyses were tested in MATLAB R2024a on 64-bit Windows. The method is intended for positive, precise decision matrices with declared benefit/cost directions, a fixed candidate set or a documented reference-boundary protocol, and explicitly stated interpretation conditions.