Exploring good cycling cities using multivariate statistics
Published in Environment systems and decisions, 2020
Some U.S. cities are excellent for cycling, like Portland, and some cities are not so good. This observation raises the question: what are the characteristics of a city that make it good for cycling? This study investigates the characteristics of 119 cities to explore what factors help make a city good for cycling. What “good” means in terms of cycling cities is subjective and we use the popular Bicycling Magazine ranking of cities for this purpose. We collected a variety of data sources about our cities including geographic, meteorology, and socioeconomic data. These data were used to conduct cluster analyses and create multivariate generalized linear regression models. We hypothesized that geographic and meteorology factors were important in determining good cycling cities. However, our hypothesis was proved wrong because socio-economic factors, like house pricing and obesity rates, play a more important role. For example, hilly cities, like San Francisco, can have excellent cycling infrastructure. The analysis shows what cities are like each other, regarding our considered characteristics; thus, city planners might wish to look at similar cities to help determine forecasts of expected use and public benefit of cycling. We use a case study of the Hampton Roads region of Virginia to show the application of our regression models.
Recommended citation: Collins, Andrew J; Jordan, Craig A; Robinson, R Michael; Cornelius, Caitlin; Gore, Ross. (2020). "Exploring good cycling cities using multivariate statistics." Environment systems and decisions. 40, 526-543.
Download Paper