Hotel Booking Cancellation Analysis

A Data Analysis course project that explores the main drivers behind hotel booking cancellations and produces actionable insights for the hospitality business.

Target variable: Is_Canceled (0 = not canceled, 1 = canceled)

---

📌 Project Goal

Hotel booking cancellations cause revenue loss and make planning harder (room allocation, staffing, forecasting). This project answers:

Which factors are most associated with booking cancellations, and how can hotels use data-driven insights to reduce cancellation risk?

---

📦 Dataset

• Dataset name: Hotel Booking Demand
• Source: Kaggle (jessemostipak / hotel-booking-demand)
• Hotels: City Hotel & Resort Hotel (Portugal)
• Time period: 2015–2017
• Raw shape: ~119k rows × 32 columns (each row = one booking)

The dataset file used in this repo is included as: hotel_bookings.csv

---

🧠 What’s Inside (Methods)

1) Data Cleaning & Preprocessing

• Handling missing values (median/mode/mean depending on the feature type)
• Removing low-quality / inconsistent records (e.g., invalid ADR values)
• Dropping highly-missing columns (e.g., Company)
• Standardizing column names and data types

2) Exploratory Data Analysis (EDA)

• Distributions and relationships for key variables such as:
  • Lead time, ADR (price), stay duration, guests
  • Market segments, customer types, seasonality

3) Feature Engineering

Examples of engineered features:

• Total_Nights = Stays_In_Weekend_Nights + Stays_In_Week_Nights
• Total_Guests = Adults + Children + Babies
• Season/long-stay indicators
• Revenue-related features

4) Feature Selection

Multiple approaches were used to identify the most relevant predictors:

• Correlation analysis
• Lasso regression
• Recursive Feature Elimination (RFE) with Logistic Regression

5) Statistical Testing

To validate whether differences are statistically meaningful:

• Normality checks
• Mann–Whitney U test for numeric features
• Chi-square test for categorical features

6) Dimensionality Reduction & Clustering

• PCA to reduce dimensionality for customer-behavior features
• K-Means clustering to group customers with similar booking behavior

---

📁 Repository Structure

.
├── Hotel_Analysis.ipynb        # Main notebook (analysis + plots + modeling steps)
├── hotel_bookings.csv          # Dataset used in the notebook
├── Documention_DA.pdf          # Full project documentation/report
└── Presentation DA.pdf         # Project presentation slides

---

🚀 Getting Started

Option A — Run locally

1. Create a virtual environment (recommended)
2. Install dependencies:

pip install numpy pandas matplotlib seaborn scipy scikit-learn jupyter

3. Launch Jupyter and open the notebook:

jupyter notebook

4. Run Hotel_Analysis.ipynb from top to bottom.

Option B — Run in Google Colab

Upload Hotel_Analysis.ipynb and hotel_bookings.csv to Colab and run the cells.

---

✅ Outputs

• Business-focused insights about cancellation drivers
• Statistical evidence of significant relationships
• Customer segments (clusters) based on booking behavior
• A full written report and a presentation deck (PDFs)

---

📝 Notes

• This repository focuses on analysis + insights + segmentation.
  If you want an end-to-end prediction model (train/test metrics), you can extend the notebook with a classification pipeline and evaluation metrics.

---

🙏 Acknowledgments

• Dataset provider: Hotel Booking Demand on Kaggle (as credited above)