mnds18/agentic-ts-forecasting-system

   

Agentic Time Series Forecasting System

This project showcases a modular, multi-agent GenAI system for time series forecasting using a retail sales use case. Designed for enterprise readiness and deployable on local machines, Docker, or cloud platforms.

🚀 Key Features

• —🧠 Agent-Oriented Architecture — Each agent (BA, DS, Reviewer, MLOps, PM, Exec) mimics real roles
• —📈 Model Variety — Prophet, ARIMA, and XGBoost (optional extensions)
• —🧪 Robust Testing — Pytest-based unit tests with CI via GitHub Actions
• —📊 Live Dashboard — Streamlit app with visual charts, error metrics, and execution logs
• —🔁 REST API — Forecast serving via local Flask
• —📦 Docker Support — Easily containerized for local/cloud deployment
• —📝 Markdown Reporting — Agent-generated reports, plans, and presentations
• —🌐 GitHub Pages — Public-facing docs, walkthroughs, and guides

💻 Quick Start

1. Clone and Set Up

git clone https://github.com/mnds18/agentic-ts-forecasting-system.git
cd agentic-ts-forecasting-system
pip install -r requirements.txt

2. Run Agentic Workflow

python agentic_ts_forecasting_system.py

3. View Streamlit Dashboard

streamlit run streamlit_app.py

4. Query Forecast via REST API

curl http://localhost:5001/forecast

🧪 Testing

Run all agent-level tests:

pytest -v test_agents.py

📦 Docker Deployment

docker build -t agentic-ts-app .
docker run -p 5001:5001 agentic-ts-app

🤗 Hugging Face Space

You can deploy this project to a Hugging Face Space:

• —Create a new space under your account
• —Set sdk: streamlit, point to streamlit_app.py
• —Add required requirements.txt and optionally Dockerfile

👉 View: https://huggingface.co/spaces/mnds18/agentic-ts-forecasting-system

📘 Documentation

• —WIKI.md — System architecture, agent logic, flow diagrams
• —CONTRIBUTING.md — Dev setup, issue flow, PR standards
• —LICENSE — MIT License

🙌 Credits

Built with ❤️ by Mrig Debsarma and contributors.

Inspired by enterprise-grade AI adoption in forecasting, orchestration, and intelligent automation.

📌 Welcome & Community

Hi there 👋 — and welcome to one of the most advanced open-source projects in agentic AI and time series forecasting!

This repo showcases an enterprise-ready, modular GenAI system using a multi-agent workflow, with live dashboards, documentation, Docker packaging, and Hugging Face deployment.

🔍 What You’ll Find Here

• —Streamlit Dashboard: 📊 Live Forecasting App
• —Hugging Face Space: 🤗 Streamlit app
• —Docs and Guides: 📘 WIKI.md, CONTRIBUTING.md
• —Agentic Workflow: 🧠 agentic_ts_forecasting_system.py
• —API Endpoint: GET /forecast → localhost:5001/forecast

🧪 How to Contribute

• —Fork the repo
• —Follow CONTRIBUTING.md
• —Run tests with pytest
• —Open your first PR 🚀

⭐ Support This Project

• —Star 🌟 the repo
• —Share it with your network
• —Submit issues for bugs, ideas, or improvements

Say hi below 👇 or request your first issue to be assigned. Let’s build the future of autonomous AI workflows together!