Optimizing Machine Learning Models for Obesity Risk Prediction through Hyperparameter Tuning

Early risk prediction is essential as obesity is a major public health concern associated with numerous chronic conditions. This study evaluates the effectiveness of various machine learning algorithms in predicting obesity risk with a particular focus on Hyperparameter optimisation enhances model performance. Obesity-related data was analysed using several algorithms including Decision Tree (DT), Support Vector Machine (SVM), K-Nearest Neighbours (KNN), and Logistic Regression (LR). Experimental results showed that Logistic Regression achieved the highest accuracy (99.63%), while Decision Tree produced the lowest (79.68%). Performance of SVM (89.6%) and KNN (94.33%) fell in between. Hyperparameter tuning significantly improved these models leading to greater robustness and predictive accuracy. Furthermore, a novel framework combining feature engineering with a cuckoo-inspired optimisation model was proposed to further enhance prediction quality. In addition, a novel hybrid framework integrating cuckoo-inspired feature optimization with Hyperparameter-tuned machine learSning models is proposed. This joint optimization strategy significantly enhances prediction accuracy, robustness, and model interpretability, distinguishing the proposed approach from existing obesity risk prediction methods.