AshimweGeoffrey/Child-malnutrition-geospatial-rwanda

Nisr Food Security Dataset # Geospatial & ML Analysis of Child Malnutrition in Rwanda Research-oriented system for mapping and predicting district-level child malnutrition risk in Rwanda. This project combines geospatial analysis, statistical profiling, and machine-learning modelling using official NISR datasets to produce actionable, district-level risk scores and interactive visualizations. Live demo - Analytics Presentation: nisrmalnutrution.vercel.app - Prediction Model: nisrmalnutrition.vercel.app What this repo does - Map spatial hotspots of child malnutrition across Rwanda’s 30 districts. - Produce district-level analytic dashboards covering nutrition, socioeconomic, and agricultural indicators. - Train and expose a Python ML model that outputs district malnutrition risk scores. - Provide frontend tools for interactive visualization and model interpretation. Research objectives - Identify spatial clusters of high malnutrition risk. - Quantify district-level predictors using cleaned and engineered features. - Build a reproducible ML workflow for district risk classification. - Deliver a visual interface to support evidence-based policy decisions. Key outputs - Hotspot Map: District-level geospatial visualization of stunting and undernutrition. - Analytics Dashboard: Nutrition, economic, and agricultural indicators with district comparisons. - ML Model: Random Forest–based pipeline with engineered composite indicators and evaluation metrics. - Interpretation Tools: Summary reports and model-driven district insights. Repository structure - ml_model/ — Training notebooks, preprocessing scripts, feature engineering, reproducible pipelines, and evaluation metrics. - Nisr-Data_analysis/ — Raw data cleaning, exploratory analysis, provenance notes, and data-preparation scripts used to derive model features. - nisr-frontend/ — React app for geospatial maps, district analytics, report exports and dashboards. - react-web-prediction_model/ — Minimal React + TypeScript demo fo …