Article Details

Abstract

This paper aims to use the XGBoost model to predict obesity levels and explain the contribution of various features to obesity risk through the SHAP method, so as to identify key influencing factors and provide a scientific basis for obesity prevention. Modeling is conducted based on multiple features such as family history of obesity, dietary habits, and frequency of physical activity. XGBoost is used to predict obesity levels, and SHAP values are applied to analyze the impact of each feature on the model output, to explain the contribution of each feature to obesity classification. Family history of obesity, age, and dietary habits are key factors affecting obesity. SHAP analysis further reveals the specific contributions and impact of these factors on obesity classification. By combining the efficient predictive ability of XGBoost and the explanatory analysis of SHAP, this research not only identifies the key features that affect obesity, but also provides a scientific basis for personalized health management and obesity prevention, demonstrating the application potential of Machine Learning in the field of public health.

Keywords

SHAP; XGBoost; Big Data; obesity level; health management

References

1. [1] Dugan T M，Mukhopadhyay S，Carroll A，et al. Machine Learning Techniques for Prediction of Early Childhood Obesity [J].Applied Clinical Informatics，2015，6（3）：506-520.
2. [2] Dirik M.Application of machine learning techniques for obesity prediction：a comparative study [J]. Journal of Complexity in Health Sciences，2023，6（2）：16–34.
3. [3] Palmieri F，Akhtar N F，Pané A，et al. Machine learning allows robust classification of visceral fat in women with obesity using common laboratory metrics [J/OL].Scientific Reports，2024，14（1）：17263[2024-08-20]. https://www.nature.com/articles/s41598-024-68269-y.
4. [4] Li Zihao, Jiang Shu Reconstruction and interpretability analysis of acoustic logging curves based on machine learning and SHAP algorithm [J]. Geological Science and Technology Bulletin, 2025, 44 (1): 321-331
5. [5] Lv Huimin, Liu Mingfeng, Jin Qianqian, etc Construction of an interpretable inference model for early injury time guided by SHAP analysis [J]. Chinese Journal of Forensic Medicine, 2024, 39 (3): 320-326
6. [6] UCI Machine Learning Repository. Estimation of Obesity Levels Based on Eating Habits and Physical Condition[R/OL].（n.d.）[2024-11-04]. https://archive.ics.uci.edu/dataset/544/estimation+of+obesity+levels+based+on+eating+habits+and+physical+condition.
7. [7] Liu Dong, Liu Ruili, Weng Haiguang Research on Text Detection Model for Cyberbullying Based on SHAP Interpretation Tool [J]. Journal of People's Public Security University of China: Natural Science Edition, 2024, 30 (3): 59-69
8. [8] Wei Hongliang, Zhang Xia, Wu Peijia, etc Research on the influencing factors of traffic capacity in construction work areas based on SHAP analysis [J]. Traffic Engineering, 2024, 24 (8): 23-30+37
9. [9] Zhang Xiaohan, Zhu Hanlin, Han Zhijiang, etc The value of SHAP value in distinguishing parotid gland malignant tumors from Warthin tumors in XGBoost MRI model [J]. Chinese Journal of Clinical Medical Imaging, 2024, 35 (7): 462-466
10. [10] Wu Jie, Chen Difang, Li Ruitong, etc Research on Personal Credit Risk Assessment Based on XGBoost and SHAP Methods [J]. Modern Information Technology, 2024, 8 (8): 146-150+155
11. [11] Zhang Ze, Chu Zhe Research on Student Achievement Prediction and SHAP Feature Analysis Based on XGBoost [J]. Information Technology Research, 2024, 50 (3): 34-40

About the authors

Huang Dongsheng (2003-), male, Han ethnicity, from Bozhou, Anhui Province, is currently pursuing a bachelor's degree with a research focus on applied statistics.