Data and Reproducibility Code for Driver-Aware Fuzzy-Kemeny Analysis of Biodiesel Industrialization Barriers in Bangladesh

This record contains anonymized analytical data and reproducibility code for a driver-aware fuzzy-Kemeny analysis of barriers to biodiesel industrialization in Bangladesh. The release consists of two files: Biodiesel_Fuzzy_Kemeny_Master_Dataset.xlsx — an anonymized master workbook containing the driver survey, barrier survey, analysis-ready barrier data, item codebooks, documented cleaning decisions, theme mappings, and canonical analytical rules. Biodiesel_Fuzzy_Kemeny_Reproducibility_Notebook.ipynb — a self-contained Google Colab notebook that reproduces the preferred-clean fuzzy-Kemeny analysis, exact Kemeny mixed-integer optimization, policy-weight scenarios, theme-level readiness gaps, synchronized analytical figures, and a 2,000-replication complete-pipeline bootstrap analysis. The driver panel contains 25 response records across 20 items. The primary barrier panel contains 20 response records across 15 items, with one additional barrier record retained for sensitivity analysis. The driver and barrier panels are separate and are not linked or matched at the respondent level. The preferred analytical specification excludes driver item D17 because of near-duplicate wording, reverse-codes barrier item B04 as 6 minus the raw response, and excludes barrier item B06 because its direction is ambiguous relative to direct barrier severity. Likert responses are represented using bounded triangular fuzzy numbers and defuzzified using the graded-mean expression (L + 4M + U) / 6. Exact Kemeny rankings are obtained through mixed-integer linear optimization. Bootstrap uncertainty is propagated through the complete integrated pipeline using 2,000 replications and random seed 20260812. Respondents are sampled independently with replacement within each panel, while all item responses belonging to a sampled respondent move together. Cleaning decisions, theme mappings, fuzzy mappings, consensus weights, and scenario weights remain fixed. Fuzzy aggregation, exact Kemeny optimization, consensus scoring, theme aggregation, readiness-gap estimation, and final driver-aware ranking are recomputed in every replication. Direct identifiers, including names, organization names, telephone numbers, email addresses, timestamps, contact details, and free-text responses, are not included. The professional-sector variable is a coarse privacy-preserving category derived from affiliation and should not be interpreted as a self-reported job title. The data represent small purposive expert panels and are not nationally representative. Theme-level findings based on limited item coverage, particularly the government-policy readiness gap, should be interpreted more strongly in terms of direction than exact magnitude. Running the notebook from top to bottom in Google Colab generates all analytical tables, figures in PNG and PDF formats, bootstrap summaries, run metadata, and a downloadable results archive.