Hotel Booking Cancellation Prediction

A machine learning pipeline designed to predict hotel booking cancellations using a Random Forest Classifier. This project emphasizes rigorous data pre-processing, feature engineering, and exploratory data analysis (EDA) to extract actionable insights for hotel management.

---

Table of Contents

1. Project Overview
2. Key Findings
3. Project Structure
4. Pipeline and Methodology
5. Installation and Setup
6. Usage Instructions

---

Project Overview

• Objective: Predict whether a hotel guest will cancel their booking based on historical booking data.
• Model Used: Random Forest Classifier.
• Performance: Achieved a predictive accuracy of 86.33%.
• Context: Developed as part of a Data-driven Artificial Intelligence module.

---

Key Findings

Through feature importance extraction, the model identified the primary drivers of booking cancellations:

1. Lead Time: Guests who book far in advance are significantly more likely to cancel.
2. Previous Cancellations: A history of cancellations is a strong predictor of future cancellations.
3. Deposit Type: The type of deposit paid (or lack thereof) heavily influences commitment to the booking.

These insights allow hotel management to identify at-risk bookings early and adjust overbooking or deposit strategies accordingly.

---

Project Structure

HOTEL-BOOKING-CANCELLATION-PREDICTION/
│
├── CODEBASE_DOCS.md                              # Detailed breakdown of codebase logic
├── hotel-booking-cancellation-prediction.ipynb   # Main executable Jupyter Notebook
├── readme.md                                     # Project documentation
├── CHS2406_Coursework1_Assignmnet_brief.docx     # Assignment brief (ignored via .gitignore)
└── hotel_bookings.csv                            # Dataset (ignored via .gitignore)