iklh-weights: Data and code for an uncertainty and sensitivity analysis of Indonesia's Environmental Quality Index

Abstract

Data and code accompanying the manuscript “What Drives Indonesia’s Environmental Quality Index? Nominal Weights, Effective Importance, and the Robustness of Provincial Rankings” (submitted to Environmental Data Science). The package audits Indonesia’s provincial Environmental Quality Index (Indeks Kualitas Lingkungan Hidup, IKLH), a weighted sum of water (0.340), air (0.428), marine water (0.099) and land (0.133) quality indices. Using the published component values for 34–38 provinces in 2021–2024, it reproduces the official index, measures each component’s effective importance (its covariance share of the between-province variance of the index) with bootstrap intervals, runs a variance-based uncertainty and sensitivity analysis of the provincial rankings over weights, normalisation, aggregation, the choice of land component and the treatment of a missing marine value (Monte Carlo with the Saltelli design and Sobol indices), assesses year-to-year rank changes with paired uncertainty intervals, and derives importance-calibrated weights that would make effective importance equal the official weights. Version 1.1.0 treats the marine component of landlocked Papua Pegunungan as structurally unavailable rather than missing, rescales min–max normalisation to [1, 100] so that geometric aggregation is defined, replaces the comparison of rank changes with rank intervals by paired intervals of the rank change, and renames corrected weights as importance-calibrated weights. Details are in CHANGELOG.md. Contents: data/ (component values by province, 2021–2024), code/ (analysis and figures), results/ (every number reported in the manuscript, rank intervals and paired rank changes), figures/, README.md, CHANGELOG.md, requirements.txt, CITATION.cff and licence files. The indicator values are official statistics from Statistics Indonesia (Badan Pusat Statistik), reproduced unchanged; please cite the original BPS table as well as this record. The same provincial data are archived with two companion studies (https://doi.org/10.5281/zenodo.23076637, https://doi.org/10.5281/zenodo.23087956). Code: MIT licence. Data and result files: CC BY 4.0.