Smart City Development Using Transit-Oriented-Development (T.O.D) and Walkability

Walkability is one of the fundamental transport components to consider when planning a new city. However, there has been a lack of research nor application on defining a detailed correlation between walkability and Transit Oriented Development (T.O.D) components, particularly when planning a new smart city. This paper aims to identify the correlation between the T.O.D components and walkability in high- and low-density cities using a machine learning algorithm. To do so, this study analyzes key explanatory variables affecting the increase or decrease of walkability. Then, a decision-tree ensemble CATBoost model is used to predict walkability using public transit data, including shared transport modes such as bicycle and electric quick scooters, supply characteristics, amenities information, street networks, and urban future network elements. By using the SHAP algorithm, the explainable A.I. method, this study seeks to determine which T.O.D component affects walkability positively or negatively.