https://github.com/sedefkjamili/dengai-ml-prediction

Machine learning project for predicting dengue fever outbreaks using climate and environmental data.
https://github.com/sedefkjamili/dengai-ml-prediction

data-science dengue gradient-boosting healthcare machine-learning python scikit-learn time-series

Last synced: about 1 month ago
JSON representation

Machine learning project for predicting dengue fever outbreaks using climate and environmental data.

• Host: GitHub
• URL: https://github.com/sedefkjamili/dengai-ml-prediction
• Owner: sedefkjamili
• License: mit
• Created: 2026-05-13T19:55:09.000Z (2 months ago)
• Default Branch: main
• Last Pushed: 2026-05-13T21:17:29.000Z (2 months ago)
• Last Synced: 2026-05-13T22:13:57.641Z (2 months ago)
• Topics: data-science, dengue, gradient-boosting, healthcare, machine-learning, python, scikit-learn, time-series
• Language: Jupyter Notebook
• Homepage:
• Size: 9.46 MB
• Stars: 0
• Watchers: 0
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# 🦟 DengAI

Machine learning project for predicting dengue fever outbreaks using climate and environmental data.

DengAI is a machine learning project focused on predicting weekly dengue fever outbreaks using climate and environmental data. The project compares traditional regression techniques with ensemble learning methods to forecast dengue case counts for two tropical cities: San Juan (Puerto Rico) and Iquitos (Peru).

## 🚀 Overview

Dengue fever is a mosquito-borne disease strongly influenced by environmental and climatic conditions such as temperature, humidity, precipitation, and vegetation. Early prediction of outbreaks can help public health organizations allocate resources more effectively and reduce the spread of disease.

This project uses machine learning models trained on historical climate and epidemiological data from the DrivenData DengAI competition dataset.

## 🔍 Features

### 📊 Data Preprocessing

- Missing value imputation using city-specific averages

- Time-aware train/validation split

- Separate modeling for each city

- Seasonal feature handling using `weekofyear`

### 🔬 Feature Engineering

- Climate and environmental indicators

- Vegetation indices (NDVI)

- Temperature and humidity variables

- Time-series aware structure

### 🧠 Machine Learning Models

- Polynomial Regression (Baseline)

- Gradient Boosting Regression

- Time-series forecasting evaluation

### 📈 Performance Evaluation

- Mean Absolute Error (MAE)

- Comparison between baseline and ensemble models

- Forecasting horizon analysis

- Feature importance visualization

## 🗂️ Repository Structure

```text

DengAI/

├── data/                  # Prediction outputs and datasets

├── notebooks/             # Jupyter notebooks for training and analysis

├── reports/               # Presentation slides and project report

├── images/                # Figures, plots, and visualizations

├── src/                   # Source code for preprocessing and modeling

├── requirements.txt       # Python dependencies

└── README.md              # Project documentation

```

## 🧪 Technologies & Libraries

- Python 3.10+

- scikit-learn

- Pandas

- NumPy

- Matplotlib

- Seaborn

- Jupyter Notebook

## ⚙️ Installation

Clone the repository:

```bash

git clone https://github.com/YOUR_USERNAME/DengAI.git

cd DengAI

```

##  Install dependencies:

```bash

pip install -r requirements.txt

🧠 Model Training & Evaluation

```

Run the notebook:

```bash

jupyter notebook

```

or execute training scripts inside /src.

## 📊 Results

Gradient Boosting Regression significantly outperformed Polynomial Regression.

Polynomial models showed overfitting at higher degrees.

Best performance was achieved using short-term forecasting horizons.

Environmental variables such as temperature and vegetation indices were among the most influential features.

## 📚 Dataset Source

- [DrivenData – DengAI: Predicting Disease Spread](https://www.drivendata.org/competitions/44/dengai-predicting-disease-spread/)

- Climate and environmental datasets for:

  - San Juan, Puerto Rico

  - Iquitos, Peru

## 👩‍💻 Authors

- Rabia Şevval Aydın

- Sedef Kjamili

Istanbul Technical University - Machine Learning Course Term Project

## 📄 License

This project is for academic and educational purposes.