Built-Environment Vulnerability and Infrastructure Deficit Dataset for Disaster Risk Reduction in Rangpur City Corporation, Bangladesh
Description
This dataset supports the article and demonstrates how an uncertainty-aware urban vulnerability index can be developed from routine municipal cadastral data in a data-scarce secondary city. It provides ward-level BVI, IDI, TSI and UPVI results for 27 RCC wards, with six excluded due to ambiguous drainage records. It includes three weighting schemes, sensitivity tests, Moran’s I/LISA statistics, Monte Carlo rank stability and construction-year uncertainty analysis. Infrastructure deficit and building fragility explain over 92% of UPVI variance, with strong spatial clustering. Confidential RCC source layers are excluded, but scripts, a reproducible six-ward sample and full documentation are included.
Files
Steps to reproduce
Steps to Reproduce: 1. Set up the environment: Use Python 3.14 and install the dependencies listed in requirements.txt. 2. Reproduce the standalone UPVI workflow: From the code directory, run build_upvi.py. This uses only the files in data/verification_sample/ and requires no confidential inputs. 3. Verify the published UPVI values: Run verify_upvi_sample.py to recompute BVI, IDI, TSI and UPVI for the six-ward verification subset using the judgement weights: BVI = 0.40, IDI = 0.35, TSI = 0.25. Compare the results with the published values. 4. Reproduce uncertainty and sensitivity analyses: Run 05_monte_carlo_uncertainty.py on data/T07_composite_index.csv for 10,000 Dirichlet draws using seed 260407. Run 06_construction_year_sensitivity.py for deterministic and 500-run multiple-imputation analysis using seed 260408. 5. Reference analyses requiring restricted data: 02_dem_topographic.py, 03_composite_index.py, and additional_analyses.py require confidential RCC source layers and are provided for pipeline transparency only. 6. Check consistency: Confirm that reproduced outputs match the manuscript’s reported ward classifications, scheme-stable wards, and Monte Carlo top-tertile probabilities.