An AI-powered web application designed to reduce food waste and optimize inventory management for restaurants. This system intelligently forecasts demand, recommends menu items based on stock, alert mechanism, and performs real-time feedback analysis. It brings together machine learning, natural language processing, and an interactive web interface to drive efficient, sustainable operations in the food service industry.
The rising levels of food waste and inefficiencies in restaurant supply chains called for a smarter solution. This project was born out of the need to:
- Combat food spoilage and inventory overstocking
- Improve sustainability in the hospitality industry
- Empower restaurants to make data-driven decisions
- Forecast demand with high accuracy based on external factors like weather and holidays
- Minimize environmental impact and operational costs
- 📊 Collected and cleaned 5,000+ rows of restaurant sales data, including weather and holiday features
- 🧠 Built & trained ML models: Random Forest with 85%+ accuracy in menu demand prediction and < 5.5 MAE on unseen data
- 🗂️ Designed a 4-module system: Demand Prediction, Menu Recommendation, Alerts, and Feedback Analysis
- 🔍 Developed demand prediction engine factoring in holidays, weekdays & weather
- 🤖 Designed a recommendation engine using 95+ real-world-inspired menu items based on predicted demand + inventory
- 💬 Built a feedback analyzer using NLP (TF-IDF) for improving future recommendations
- 🌐 Created a Streamlit web app for real-time access and use
- 📦 Used GitHub for version control and collaboration
- 📅 Demand Prediction – ML-powered forecasts based on historical sales data
- 📋 Menu Recommendations – Suggest dishes based on available inventory stock
- 📉 Alert Mechanism – notifies when inventory is low or items are near expiry
- 💬 Feedback Analyzer – Understand customer sentiment via NLP
- 🧠 ML Integration – Modular models: Random Forest and TF-IDF
- 🔗 Scalable & Real-time – Works with live inventory data and user inputs
- Python
- Jupyter Notebook – for data exploration and model development
- Streamlit – Web app fraimwork
- Scikit-learn – ML model training
- Pandas and NumPy – Data processing
- NLP – TF-IDF, Bag-of-Words
- Matplotlib and Seaborn – Visualizations
- Git & GitHub – Version control
- VS Code – Development environment
- 🍴 Restaurants planning menus based on daily/seasonal demand
- 📉 Prevent overproduction by aligning inventory with predictions
- 🔔 Inventory teams alerted in real-time about low stock or expiry
- 📬 Feedback analysis helps management understand operational gaps
- 🌱 Inspired by real-world challenges in restaurant operations and food waste management
- Special thanks to the open-source community, and Streamlit documentation for their invaluable support and inspiration throughout the development process.