🏨 Hotel Reservation Cancellation Prediction - MLOps Project

This MLOps project predicts whether a hotel reservation will be canceled based on customer and booking features. It includes a full ML pipeline from data ingestion to deployment using CI/CD practices and Docker, and a user-friendly web interface built with Flask.

🌐 Live Demo: Click here to try it!

• Note: May be this link will not work in future.

---

📌 Features

• 📊 Predicts if a reservation will be canceled
• ⚙️ End-to-end MLOps pipeline
• 🐳 Dockerized for containerized deployment
• 🧪 ML pipeline includes ingestion, preprocessing, training
• 📦 Integrated with Jenkins for CI/CD
• 🎯 Deployed on Google Cloud Run
• 🧠 Uses LightGBM and MLFlow for tracking

---

🧾 Tech Stack

Layer	Tools Used
ML & Modeling	Scikit-learn, LightGBM, imbalanced-learn
Tracking	MLFlow
Web App	Flask + HTML/CSS
CI/CD	Jenkins, GitHub
Containerization	Docker
Cloud	Google Cloud Run

---

🚀 Web App Overview

The frontend allows users to enter details like:

• Lead time
• Special requests
• Room price
• Arrival month/date
• Meal plan, room type, etc.

🔮 After clicking "Predict", the app displays whether the reservation is likely to be canceled or not.

UI Screenshot

---

🧠 Machine Learning Pipeline

📂 Code Modules

.
├── src/
│   ├── data_ingestion.py       # Downloads/splits data
│   ├── data_preprocessing.py   # Cleans and prepares features
│   ├── model_training.py       # Trains and saves model
│   └── custom_exception.py     # Custom error handling
├── pipeline/
│   └── training_pipeline.py    # Runs end-to-end pipeline
├── notebook/
│   └── notebook.ipynb          # EDA and experiments

🔄 Pipeline Steps

1.Data Ingestion:

• Loads training/test data (e.g., from GCS or local)

2.Preprocessing:

• Encodes categorical features

• Scales numerical data

• Handles missing values and imbalance

3.Model Training:

• Trains a LightGBM model

• Evaluates on test data

• Logs metrics to MLFlow

• Saves the model using joblib

💻 Web Application – application.py

The Flask app:

• Loads the trained model

• Takes user input from a form

• Preprocesses and feeds it to the model

• Returns prediction result in the UI

Prediction values:

• ✅ "The Customer is not going to cancel Reservation in future."
• ❌ "The Customer is likely to cancel Reservation."