The built environment is a significant contributor to obesity, but previous studies have focused more on the built environment at residential areas, ignoring that at other geographical contexts. This study introduces an innovative framework to explore the relationships between the built environment at various geographic contexts and obesity, including residences, commuting routes, and workplaces. Based on a sample consisting of 1043 adults in Shanghai, the gradient boosting decision trees model is employed to examine how built environments around residences, commuting routes, and workplaces affect obesity, controlling for sociodemographics and air pollutants. Results show that the built environment around commuting routes makes the largest contributions to obesity, accounting for 21 %. Built environments around workplaces and residences have similar contributions to obesity, each accounted for 16 %. Moreover, some built environment elements around commuting routes and residences (population density, restaurant density, and intersection density) have similar and positive correlations with obesity. Residential and workplace transit stop densities have different associations with obesity, which are U-shaped and negative, respectively. Additionally, greenspace around commuting routes, PM2.5 and mixed land use surrounding workplaces, and NO2 surrounding residences are positively related to obesity. These findings encourage urban planners to meticulously optimize and design the built environment for reducing obesity and bolstering healthy cities.