Supporting data for "Navigating in a Layered City: Evaluating the Perceived Wayfinding Environment Qualities of Elevated Walkway Systems Using Street View Images and Deep Learning"

This dataset supports Jiang et al. (2026) published in Building and Environment. It contains the extracted streetscape features, aggregated perception ratings, predicted perception scores, space syntax metrics, and trained Random Forest classifiers used to evaluate perceived wayfinding quality across Hong Kong's elevated walkway network. The deposit allows readers to reproduce the model predictions, regenerate the spatial maps in Figures 8–11, and re-run the SHAP feature-attribution analysis. See README.md for a complete file index and data_dictionary.md for column-by-column documentation.