Urban Analytics: Sentiment, Mobility & Urban Behavior

This project collection explores how computational methods can reveal relationships between public perception, human behavior, and urban environments. The primary case study, completed with a three-person team and recognized as a top project in the course, examined whether positive sentiment toward Atlanta parks is associated with greater visitation to nearby businesses and other points of interest. We combined park boundaries, mobility-based POI visitation data, Google Maps reviews, Census indicators, transportation access, built-environment characteristics, and seasonal observations to study how perceptions of public space may relate to surrounding urban activity. The project progressed from spatial sampling and web-based review collection to NLP sentiment classification and multivariate statistical modeling. The final R workflow used negative binomial regression, model comparison, seasonal analysis, and industry-specific models while controlling for factors including Walk Score, transit access, crime, street-network characteristics, household income, population and job density, poverty, demographics, and park size. A second independent case study applies R-based text and sentiment analysis to online discourse about women serving in the U.S. military, demonstrating how similar computational methods can be applied to broader social and policy questions.