🚀 Anomaly Detection for Corporate Fleets

A machine learning system for real-time anomaly detection in fleet data using autoencoders and forecasting models.

📌 Overview

This project detects anomalies in streaming vehicle data, leveraging unsupervised learning techniques. The system is designed to work within the GeoSat infrastructure, analyzing GPS and sensor data from corporate fleets.

🏗️ Architecture

The system combines two approaches for anomaly detection:

• Autoencoder – Learns a compressed representation of normal data and detects anomalies via reconstruction error.
• Forecasting Model (coming soon) – Predicts future values and detects anomalies via prediction error.

📊 Data Processing

• Input: Rolling windows of 20 time steps, with 4 temporal features + 10 vehicle features.
• Preprocessing: Standardization (Z-score), feature extraction, and temporal windowing.

🏢 Repository Structure

• anomaly-detection-ts/
  • │── data/ # Dataset and preprocessing scripts
  • │── models/ # Machine learning models
  • │── notebooks/ # Jupyter Notebooks for analysis
  • │── utils/ # Helper functions
  • │── config/ # Configuration files
  • │── scripts/ # Execution scripts
  • │── tests/ # Unit tests
  • │── requirements.txt # Dependencies
  • │── README.md # Documentation
  • │── .gitignore # Ignore unnecessary files

🚀 Quick Start

git clone https://github.com/your-username/anomaly-detection-ts.git
cd anomaly-detection-ts
pip install -r requirements.txt
python models/train_autoencoder.py

📄 License

This project is licensed under the Creative Commons Attribution-NonCommercial 4.0 International License.
You can read the full license here.