Walkability is a hot topic for variety of disciplines, as well as everyday walker. It affects the health, the environment and the liveliness of our neighbourhoods. Walkable streets are necessary for a better lifestyle and sustainable planet. The problem with walkability is that we still don’t have a general understanding of the concept. Every study differs in the way they define walkability, thus making walkability a subjective topic. However, the subjectivity causes contradiction in science. In this study, the aim to answer the question of what makes a street walkable by using a data analytic approach. The features used in other studies, as well as new attributes specific to this study, were investigated. Street images were used to extract data. The data was divided into nine categories: Street, Sidewalk, Obstacles, Urban Blocks, Amenities, Transportation, Attractiveness, People, and Vehicles. Data collection was carried out by measuring physical attributes through Remote Sensing images in QGIS, visually analyzing qualitative attributes with Google Street Maps/View and double checking data in Open Street Map Overpass Turbo API. Attributes were translated into scores and normalized where possible. Mutual Information Matrix and Correlation processes were conducted in Rapidminer. The attributes were processed in relation to overall assessment of walkability which was defined with personal rating. As a result, Mutual Information and Correlation matrices are useful in figuring out the relationship and dependencies between different attributes. Applying data analytics to a more comprehensive dataset will help identify the global factors of walkability.
|Publication Date||December 23, 2022|
|Published in Issue||Year 2022 - Vol.23 - 16th DDAS (MSTAS) Special Issue -2022|