Machine learning projects focused on predicting categorical outcomes from real-world data.
What Classification Predicts
Classification models estimate which category an observation belongs to. In portfolio projects, this can mean predicting whether a booking will be canceled, whether a customer may churn, whether a transaction is risky, or whether a record belongs to one business segment instead of another.
My classification projects focus on clear problem framing, careful feature availability checks, class imbalance handling, model evaluation beyond accuracy, and interpretable conclusions that connect model results back to business decisions.
Hotel Booking Cancellation Prediction
Classification project focused on predicting whether a hotel booking will be canceled using booking behavior, customer context, deposit policy, and stay characteristics. The final Random Forest pipeline uses preprocessing, SMOTE inside cross-validation, hyperparameter tuning, model evaluation, and feature-importance interpretation.
Final result: ROC-AUC 0.90, accuracy 0.85, and canceled-class recall 0.71.