ADS Capstone Chronicles Revised

32

performance was validated not only through metrics like precision, recall, and R² score, but also through its interpretability and reliability, as illustrated in the following visualizations. 5.5.1 Confusion Matrix The confusion matrix for menu recommendations in Figure 20 demonstrates XGBoost’s exceptional classification performance. It shows over 30,000 correct Figure 20 Confusion Matrix for Menu Recommendations

classifications for each class and minimal misclassifications, resulting in near-perfect precision and recall. Likewise, the confusion matrix for individual food recommendations in Figure 21 illustrates similarly high accuracy with minimal false positives and negatives.

234