Supplementary Material and Reproducibility Notebooks for "Multi-Hazard Risk Assessment Framework for Rural Schools in Northwestern Colombia: Integrating App-Based Surveys, Expert Weighting, and Machine Learning"

This repository contains the supplementary tables and reproducibility notebooks associated with the article “Multi-Hazard Risk Assessment Framework for Rural Schools in Northwestern Colombia: Integrating App-Based Surveys, Expert Weighting, and Machine Learning,” published in the International Journal of Disaster Risk Reduction. Gallego-Montoya, J.J., Bonett Díaz, R.L., Rendón-Rivera, A., Blandón Uribe, C.A., & Frischknecht, C. (2026). Multi-Hazard Risk Assessment Framework for Rural Schools in Northwestern Colombia: Integrating App-Based Surveys, Expert Weighting, and Machine Learning. International Journal of Disaster Risk Reduction. https://doi.org/10.1016/j.ijdrr.2026.106221 The repository includes anonymized datasets, supplementary tables, and Jupyter notebooks used to reproduce the analyses presented in the study. These materials support the calculation of the Analytic Hierarchy Process (AHP) weighting, formula-based risk indices, Random Forest (RF) models, multicollinearity diagnostics, and the derivation of single-hazard and multi-hazard risk indices for 100 rural schools in the Department of Antioquia, northwestern Colombia. Supplementary tables included: Table S1. Data sources and acquisition variables.Table S2. Anonymized raw data for selected schools collected through the SRS mobile application.Table S3. AHP pairwise comparison matrices used for risk component weighting.Table S4. Variable standardization and AHP-based risk component indices.Table S5. Formula-based risk indices.Table S6. Predicted risk scores from Random Forest models (RF1 and RF2).Table S7. Variable importance and multicollinearity diagnostics (VIF and tolerance) for RF models.Table S8. Single-hazard and multi-hazard risk indices. Jupyter notebooks included: Notebook 01. Random Forest models RF1-RF2 and multicollinearity diagnostics.ipynbNotebook 02. Single and multi-hazard risk calculation.ipynb These data support the reproducibility of the proposed methodology and can serve as a reference for similar risk assessment frameworks in rural or data-scarce regions. This research was conducted as part of the Safe Rural Schools Project: A Multi-Hazard Risk Assessment and Reduction Technology Strategy, funded by the G8 University Alliance of the Department of Antioquia in Colombia, led by the Universidad de Medellín, the Universidad Nacional de Colombia, sede Medellín, and the Universidad EIA, under the Research, Development and Innovation (R & D + i) Projects within the Regional Agenda.