Delays in providing humanitarian aid often stem from a lack of sufficient information about the quantity of aid required for disaster-stricken regions. Responding effectively to disasters begins with understanding the needs of affected communities. This research develops a predictive model to forecast aid requirements using historical data from 2002 to 2022. By integrating multiple datasets—EM-DAT (Emergency Events Database) for disaster statistics, GLIDE (Global Unique Disaster Identifier) and ReliefWeb for event tracking, and OECD (Organization for Economic Co-operation and Development) for aid-related information, we analyze how aid is distributed across different contexts. We examined key features such as population size, disaster frequency, and temporal factors (e.g., months and dates) to assess their impact on changes in aid allocation. Our best-performing model, XGBoost optimized with GridSearchCV, achieved an R² score of 0.86, indicating high predictive accuracy. This study provides a data-driven framework to improve the efficiency of aid allocation, offering decision-makers reliable estimates to prioritize timely and appropriate responses. |
*** Title, author list and abstract as submitted during Camera-Ready version delivery. Small changes that may have occurred during processing by Springer may not appear in this window.