Check out our prior work!
Text Mining and Spatial Analysis of Yelp Data to Support Socially Vibrant Cities by Alexander Bendeck and Clio Andris
Published in Urban Computing
We used Yelp online reviews of places of interest to search for social relationship words (like mother, father, partner, boyfriend, girlfriend, friend, and coworker, for example) in eight US cities. We then mapped where in a city certain relationships spend time and found that certain places of interest (restaurants, shopping, etc.) support different types of relationships.
The map to the right shows a comparison of hot spots, or places with a significant number of Yelp reviews, specific to romantic and familial relationships.
In addition, we also made an interactive online tool that lets users select a relationship type of interest (such as "romantic" or "family") and search for places whose reviews mention these relationships. This tool can be found here.
Published in the Journal of Urbanism
We used geographic information system (GIS) and surveys of people in romantic relationships to locate where couples spend time in State College, Pennsylvania. We found that pedestrian and transportation infrastructure as well as a variety of close and affordable activities are particularly important for couples.
The map to the left shows the home locations of survey respondents, partners who live together and partners who live apart, and the top ten places of interest (POIs) such as the movie theater or certain restaurants.
Lessons From a Human-In-The-Loop Machine Learning Approach for Planning: Identifying Vacant, Abandoned, and Disinvested Properties in Savannah, Georgia
by Xiaofan Liang, Brian Brainerd, Tara Hicks, Clio Andris
Publication in progress
In this project, we used a human-in-the-loop machine learning model (a tool that combines human insights and statistics) to identify vacant, abandoned, and disinvested (VAD) properties in Savannah, Georgia. This method revealed that tax delinquency, code violations, and crime history were the best indicators of VAD properties.
The map to the right shows the difference between VAD properties as identified by the model versus identified by human judgement.
Place Social Diversity: A New Attribute for Points of Interest
by Seolha Lee and Clio Andris
Points of Interest (POIs) provide opportunities for social interaction among different social groups, which is crucial for cultivating strong and healthy communities. In this study, we create a Place Diversity Index (PDI) to measure the social diversity of visitors at POIs using trip data from mobile GPS traces. Using this index, we analyze what makes a destination socially diverse through a case study in Atlanta, Georgia. The result shows that in Atlanta, POIs are largely segregated, yet food places are most socially diverse. Also, socially diverse POIs are located in higher-income, more white neighborhoods, and areas expe riencing advanced gentrification. The method we explore in this study, which annotates a place with the socio-demographic heterogeneity of visitors, can be generalized into other cities and be useful for practitioners and researchers who want to promote social diversity in places through place-based interventions.
Paper presented at the 18th International Conference on Computational Urban Planning and Urban Management (#CUPUM) this summer. View the paper. Explore the companion interactive viewer here.
