Examining non-linear associations between population density and waist-hip ratio: An application of gradient boosting decision trees

《Examining non-linear associations between population density and waist-hip ratio: An application of gradient boosting decision trees》

打印

关键字: Compact development;Obesity;Machine learning;Threshold effects;China Labor-force Dynamics Survey

作者单位: Research Center for China Administrative Division, East China Normal University, Shanghai 200241, People's Republic of China;Institute of Eco-Chongming, Shanghai 202162, People's Republic of China;Future City Lab, East China Normal University, Shanghai 200241, People's Republic of China;Humphrey School of Public Affairs, University of Minnesota, 301 19th Ave. S, Minneapolis, MN 55455, USA;Research Center for China Administrative Division, East China Normal University, Shanghai 200241, People's Republic of China;Institute of Eco-Chongming, Shanghai 202162, People's Republic of China;Future City Lab, East China Normal University, Shanghai 200241, People's Republic of China;Humphrey School of Public Affairs, University of Minnesota, 301 19th Ave. S, Minneapolis, MN 55455, USA

摘要: Many studies have explored the relationship between population density and obesity, but there is no consensus, particularly in dense Chinese cities. This study applied gradient boosting decision trees to 2014 national survey data to examine the non-linear or threshold effects of population density at both local and regional levels on waist-hip ratio (WHR), controlling for other built environment elements and socio-demographics. Built environment elements collectively have a stronger predictive power than socio-demographics (56.6% vs. 43.4%). Within a certain range, regional population density is negatively associated with WHR, but its marginal effect diminishes beyond the upper threshold. Local population density has a U-shaped relationship with WHR. These results suggest that urban planners can alleviate the risk of obesity through population densification, but over-densification tends to be inefficient, and sometimes counterproductive.