📈 TimeSeries Anomaly Transformer

TimeSeries Anomaly Transformer is a full-stack AI system designed to detect anomalies in time-series data using a custom-built Transformer architecture. Users can upload CSV datasets, visualize anomalies, and explore detected abnormal points through an interactive UI.

---

🚀 Features

• 🤖 Transformer-based Anomaly Detection
• 🧠 Association Discrepancy + Anomaly Attention Mechanism
• 📊 Real-time Graph Visualization
• 📁 CSV Dataset Upload Support
• 🔍 Highlights Exact Anomaly Locations
• ⚡ Backend-powered anomaly scoring
• 🌐 React-based modern UI

---

🧩 Tech Stack

Layer	Technology
Frontend	React, Chart.js
Backend	Express.js
Model	Python (NumPy, Matplotlib, PyTorch)
Core Logic	Transformer, Gaussian Kernel, Minimax Strategy
Dataset	Bitcoin (2018–2024) CSV

---

📁 Folder Structure

TimeSeries-Transformer/
├── backend/                # Node + Express API
│   ├── server.js
│   ├── routes/
│   └── controllers/
│
├── model/                  # Python Transformer Models
│   ├── simple_transformer.py
│   ├── pytorch_transformer.py
│   ├── transformer_no_libs.py
│   ├── anomaly_attention.py
│   └── preprocess.py
│
├── frontend/               # React UI
│   ├── src/
│   │   ├── components/
│   │   │   ├── FileUploader.js
│   │   │   ├── GraphView.js
│   │   │   └── AnomalyList.js
│   │   ├── App.js
│   │   └── index.js
│   ├── public/
│   └── package.json
│
├── datasets/
│   ├── BTC_1D.csv
│   ├── BTC_4H.csv
│   ├── BTC_15M.csv
│   └── BTC_1H.csv
│
└── README.md

---

🔧 Setup Instructions

🐍 Python Model Setup

cd model
python -m venv venv
venv\Scripts\activate     # Windows
pip install -r requirements.txt

Run model:

python simple_transformer.py

---

🌐 Frontend (React)

cd frontend
npm install
npm start

---

🔌 Backend (Express)

cd backend
npm install
npm start

---

👥 Contributors

@koustub1412 Team Members — NGIT