Beyond Schelling's Threshold: Quantifying Emotional Impetus of Human Residential Distribution Formation through Geo-tagged Facial Analysis

Some researchers have noticed the underlying influence of subjective perception on human mobility and population distribution. For example, Schelling's segregation model incorporates settlement preferences of individuals and their relationship with population distribution. However, empirical investigations of related theories at large spatial scales remain limited due to the lack of suitable datasets. To bridge this gap, our study proposes a deep-learning based framework to obtain emotional expression records with high spatial and temporal resolution from geo-tagged Flickr photographs. Taking the United States as a case study, this research mainly focuses on: (1) the spatial-temporal trends in collective emotional expressions and population segregation pattern; (2) the potential associations between emotional states and migration dynamics.