Machine learning model to predict vaccine dropout risk (DTP1–DTP3) across African countries using immunization and demographic data.
# Vax Track AI - Vaccine Dropout Prediction
This project develops a machine learning model to predict vaccine dropout risk (DTP1–DTP3) across African countries using immunization and demographic data.
It also includes an interactive **Streamlit app** for real-time predictions.
Live Deployment: Streamlit App
---
## 🚀 Features
- Preprocessing pipeline with scaling, encoding, and SMOTE for class balance
- Models trained with **Logistic Regression, Random Forest, SVM, and XGBoost** using GridSearchCV
- Automatic region selection when a user chooses a country
- Streamlit dashboard with metric cards (**Dropout Rate, Coverage Average, DTP3 Coverage**)
- Visualizations including boxplots and interactive charts
---
## 📂 Project Structure
```bash
├── data/ # Raw and processed datasets
├── notebooks/ # Jupyter notebooks for exploration & training
├── app/ # Streamlit app
│ ├── app.py
│ └── utils.py
├── models/ # Saved models and preprocessing pipelines
├── requirements.txt # Python dependencies
└── README.md # Project documentation
```
## ⚙️ Installation
Clone this repo:
```bash
git clone
github.com
cd vaccine-dropout-prediction
```
Create and activate a virtual environment:
```bash
python -m venv venv
source venv/bin/activate # On Windows: venv\Scripts\activate
```
Install dependencies:
```bash
pip install -r requirements.txt
```
## ▶️ Usage
### Run Streamlit app
```bash
streamlit run app/app.py
```
## 📊 Example App View
- Metric cards for Dropout Rate, Coverage Average, DTP3 Coverage
- Country & region selection with automatic region mapping
- Prediction probability shown with visual feedback
## 📌 Future Improvements
- Deploy on Streamlit Cloud or Azure Web App
- Add additional health & socioeconomic predictors
- Expand dataset beyond Africa