Profiling Spatiotemporal Contexts Associated with Compound Exposure Disadvantages in Noise and Greenery: A Study Using Mobile Sensors and Explainable Machine Learning
Examines compound disadvantages from noise and greenery exposure in daily mobility, suggesting socioeconomic factors influence these outcomes.
Key Points
The aim is to explore the nonlinear relationships between spatiotemporal contexts and compound disadvantages from noise and greenery exposure using mobile sensors.
Utilized mobile sensors, GPS, satellite imagery, and activity diaries to gather data on noise and greenery exposures.
Developed a compound disadvantage index based on daily mobility patterns identified through stay and move event sequences.
Applied a novel framework incorporating GPBoost, residual-based bootstraps, and SHAP to analyze data.
Investigated sociodemographic factors such as socioeconomic and employment status in relation to compound disadvantages.
Observed the neighborhood effect averaging problem in noise and greenery measurements.
Identified socioeconomic status and spatiotemporal contexts as significant predictors of compound disadvantages.
Found employment status to be the primary socioeconomic predictor of compound disadvantages.
Bootstrapped SHAP methodology highlighted key spatiotemporal contexts and interactions with socioeconomic status.