bucky-ops/Kenya-Childhood-Malnutrition-Risk-Prediction-System
Domain:
healthcareRecord type:
software
Creator:
buc
Host:
An open-source Digital Tool for predicting acute childhood malnutrition risk in Kenya using machine learning and WHO Data Quality Review standards. This system provides county and sub-county level predictions of acute childhood malnutrition cases one month ahead, incorporating rigorous data quality monitoring and automated governance features.
# 📊 Kenya Childhood Malnutrition Risk Prediction System
> An open-source Digital Public Good for predicting acute childhood malnutrition risk in Kenya using machine learning and WHO Data Quality Review standards.
## Table of Contents
- Features
- Quick-Start
- Installation
- Usage
- Development
- Contributing
- License
## Features
- Machine Learning Predictions using Random Forest model trained on WHO and UNICEF indicators
- Data Quality Monitoring with WHO DQR-compliant validation and automated scoring
- Quality-Weighted Training that prioritizes high-quality data sources
- Automated Reporting with monthly PDF reports for stakeholders
- Alert System for email notifications about data quality deterioration
- Interactive Dashboards with Streamlit apps for data exploration and quality review
- Scenario Analysis for policy impact simulations
- Uncertainty Quantification for confidence intervals in predictions
## Quick-Start
```bash
# clone & cd
git clone github.com
cd your-repo
# create a virtual environment
python -m venv .venv && source .venv/bin/activate
# install the package and dev tools
pip install -e .[dev]
# run the main app
python -m your_pkg.app.app # or any other entry-point
# run the complete pipeline with test data
python system_launcher.py --pipeline
```
## Installation
```bash
# Create virtual environment
python -m venv venv
source venv/bin/activate # On Windows: venv\Scripts\activate
# Install dependencies
pip install -r requirements.txt
# Install as editable package (optional)
pip install -e .
```
## Usage
The system consists of several interconnected modules:
### 1. Data Pipeline
- `src.data.load_data`: Load and merge data from WHO, UNICEF, and DHIS2 sources
- `src.data.clean_data`: Clean missing values and impute data
- `src.features.build_features`: Engineer features with lagged indicators and seasonal components
### 2. Model Training
- `src.models.train_model`: Train Random Forest model with time …