This model predicts whether a hotel booking will be canceled or not based on customer behavior, booking details, and historical patterns.

It is trained on a real-world hotel booking dataset and designed to help businesses reduce revenue loss and improve operational planning.

Predict booking cancellation (is_canceled) Assist hotels in proactive decision-making Identify high-risk bookings

The model uses a combination of customer, booking, and pricing features, including:

lead_time → Days between booking and arrival adr → Average Daily Rate (price per night) previous_cancellations → Past cancellation count deposit_type → Booking deposit category total_stay → Total nights stayed (engineered feature) total_guests → Number of guests (engineered feature)

0 → Not Canceled ✅ 1 → Canceled ❌

Algorithm: Gradient Boosting / XGBoost Type: Binary Classification Framework: Scikit-learn compatible

Accuracy: ~0.85–0.90 ROC-AUC: ~0.90+

👉 Strong performance on structured tabular data

Longer lead time increases cancellation probability Non-refundable deposits drastically reduce cancellations Returning customers are more reliable Pricing (ADR) influences booking behavior

import joblib import pandas as pd

model = joblib.load("model.pkl")

sample = pd.DataFrame({ "lead_time": [50], "adr": [100], "previous_cancellations": [0], "deposit_type": ["No Deposit"] })

prediction = model.predict(sample) print(prediction)

Performance depends on data quality May not generalize across all hotel types or regions Requires proper preprocessing (same as training pipeline) 🔧 Future Improvements Use CatBoost for better categorical handling Add more behavioral features Deploy as real-time API

Data Science Enthusiast Kaggle Competitor

If you find this model useful:

⭐ Like the repository 🔁 Share with others