An adaptation of reference class forecasting for the assessment of large-scale urban planning vision, a SEM-ANN approach to the case of Hong Kong Lantau tomorrow

Abstract

An outline strategic urban planning vision is a front-end government strategic plan with sparsely defined goals. Methods for an ex-ante appraisal of such sparsely defined vision are limited in the literature. By adapting a reference class forecasting (RCF) methodology, we propose an innovative two-stage combination of Structural Equation Modeling (SEM) and Artificial Neural Networks (ANN) as an explainable ANN strategy to the appraisal of urban planning vision outline. The SEM-ANN operationalizes interaction between job, resident, and multi-modal accessibility in a public transport-dominated city. This strategy is applied to Lantau Tomorrow Vision in Hong Kong, as an extreme case study of a large, reclaimed island. The vision is broadly outlined as a New Town with a third CBD, residential and job targets, road and urban rail transport infrastructure routing, and an overall cost. The results show that the New Town scenario job/population goal should be plausibly attainable by increasing the transport infrastructure accessibility supply. Yet, the simulation indicates that the CBD3 employment goal based on CBD1 is out of range. Overall, our SEM-ANN method, as an adaptation of RCF, is of particular interest in front-end large-scale outline urban planning vision appraisal.