Nyando Basin Flood Risk Prediction System: A Production-Grade, Explainable Machine Learning Framework for Ward-Level Flood Susceptibility Mapping in Kisumu County, Kenya
Year: 2026
Type: article
Abstract: A production-grade machine learning flood susceptibility system for the Nyando River Basin, Kisumu County, Kenya. XGBoost classifier achieving AUC-ROC 0.94 trained on open satellite data (CHIRPS, NASA DEM, Sentinel-1 SAR). Includes modular Python source package, 19 pytest unit tests, FastAPI prediction endpoint, and React web dashboard. 100% open data — Kenya DPA 2019 and GDPR compliant. Full codebase, training data, model weights, and unit tests published under MIT and CC-BY-4.0 at github.com/jameskoero/nyando-flood-ai.
Author: Koero, James
Language: English
Subfield: Global and Planetary Change
Field: Environmental Science
Domain: Physical Sciences
Open Access status: gold
OpenAlex ID: W7160613314
FWCI:
Cites:
Cited by:
Related to: