Derio001/Chad-malnutrition-prediction

ML pipeline predicting child malnutrition risk in Chad using DHS 2014 survey data. Gradient Boosting achieved 92% accuracy and 0.979 AUC on 9,826 children. 52.9% of Chadian children under five are malnourished. # 🍽️ Child Malnutrition Risk Prediction in Chad > **Over half of Chadian children under five are malnourished.** > This project builds a machine learning pipeline to predict which > children are at highest risk — using 9,826 children from the > DHS Chad 2014 survey — enabling NGOs and health workers to > intervene before malnutrition becomes acute or fatal. --- ## 🌐 Live Demo 👉 **Try the app here** Enter a child's age, weight, and height to get an instant malnutrition risk assessment — designed for community health workers and NGO field teams in Chad. --- ## 🌍 Why This Matters Chad has one of the highest child malnutrition rates in the world: - **42.9%** of children under five are stunted - **32.5%** are underweight - **14.2%** are wasted — above the WHO emergency threshold - **52.9%** suffer from some form of malnutrition Yet most humanitarian responses are **reactive** — children are identified only when they arrive at a health facility in crisis. This project asks: **can we predict which children are at highest risk before they become acutely malnourished?** --- ## 📊 Dataset - **Source**: DHS Program — Chad Standard DHS 2014 - **File**: Children's Recode (KR) — 18,623 total children - **Measured**: 9,826 children with anthropometric measurements - **Target**: Any malnutrition (stunting OR underweight OR wasting) ### Malnutrition Definitions (WHO Standards) | Indicator | Column | Threshold | Prevalence | |-----------|--------|-----------|------------| | Stunting | HW70 (HAZ) | Z-score **Note on data**: DHS data requires free registration at > dhsprogram.com. Request Chad 2014 > KR recode dataset. The processed modeling dataset is included > in this repo. --- ## 📁 Project Structure ``` chad-malnutrition-prediction/ │ ├── Chad_Malnutrition_Prediction.ipynb # Full pipeline ├── chad_malnutrition_model_data.csv # Processed dataset ├── malnutrition_model_comparison.csv # Model results ├── malnutrition_feature_importance.csv # Featu …