ADS Capstone Chronicles Revised

31

Table 7 Best Regression Results

0.995479, with similarly high values for accuracy, recall, and F1 score. This near-perfect performance demonstrates XGBoost’s ability to handle structured data with minimal variability, providing reliable classifications for menu items. For individual food recommendations, XGBoost again led the results, achieving a precision of 0.988664 and an F1 score of 0.989270, as shown in Table 8. The Random Forest Classifier followed closely, with a precision of 0.982445 and an F1 score of 0.980483. While Random Forest performed well, XGBoost’s slightly higher precision and recall solidified its position as the preferred model for both datasets.

5.4 Grid Search Optimization for Classification Models Grid search optimization also refined the classification models, focusing on XGBoost Classifier and Random Forest Classifier as the top contenders from the baseline results. Metrics such as accuracy, precision, recall, and F1 score were calculated to assess the models’ ability to classify menu and individual food items as “recommended” or “not recommended.” For both datasets, the XGBoost Classifier achieved superior performance. As seen in Table 8, the precision for menu recommendations reached an outstanding

Table 8 Best Classification Model Results

model based on its exceptional performance metrics, robust classification accuracy, and interpretability. Its capability to handle both structured and semi-structured data makes it an excellent choice for providing reliable and actionable dietary recommendations tailored to diabetic patients. XGBoost's superior

5.5 Final Model Selection: XGBoost as the Best Overall Model The final evaluation results demonstrate that the XGBoost Classifier consistently outperformed all other models across both datasets for regression and classification tasks. The XGBoost Classifier was chosen as the final

233