Hotel Booking Cancellation Prediction App 🏨🚫

📅 A Flask-based predictive app that analyzes hotel booking details to predict cancellation probability. Features optimization improved model accuracy by 8% 📈.

Try the app! 📱

You can try the it live by clicking here.

Note: The app is deployed on a platform where the deployment includes a limitation of 10 free sessions. If the app is not working, it's likely because the session limit has been reached.

Features 🔧

• Predicts hotel booking cancellation likelihood.
• User-friendly web form for inputting key features.
• Provides real-time cancellation prediction & probability.
• Explored and Pruned Random Forest, LightGBM for Recall Optimization
• Randomized Search adn Optuna for Hyperparameter Tuning:
• Dynamically Adjusting Threshold to Visualize Performance

Installation 📝

1. Clone the repo:

   git clone https://github.com/amerob/ReservePredict.git
   cd ReservePredict

2. Set up virtual environment:

   python -m venv venv
   source venv/bin/activate  # Windows: `venv\Scripts\activate`

3. Install dependencies:

   pip install -r requirements.txt

4. Add your predict_pipeline.pickle file to the root directory.

Usage 🚀

1. Run the app:

   python app.py

2. Visit http://127.0.0.1:5000/ in your browser.

3. Input features:

   • Type of Meal Plan 🍽️
   • Room Type Reserved 🛏️
   • Market Segment Type 📊
   • Lead Time 🕰️
   • Average Price per Room 💵
   • Number of Special Requests ✨

4. Get your result: The app shows the cancellation prediction and probability!

File Structure 📂

hotel-booking-cancellation-prediction/
│
├── app.py                    # Flask app
├── predict_pipeline.pickle   # Trained model
├── templates/
│   └── index.html            # Frontend HTML
├── requirements.txt          # Python dependencies
├── Aptfile                   # System dependencies for deployment
├── Procfile                  # Specifies how to run the app (used by platforms like railway and Heroku)
└── README.md                 # This file

How It Works ⚙️

1. Model: The model predicts cancellation probability based on input features.
2. Input: Users input data via the form (meal plan, room type, etc.).
3. Prediction: The app processes the input, feeds it into the model, and returns the cancellation likelihood and probability.

License 📄

MIT License - see the LICENSE file for details.