This study “mined” the Auckland Transport Active Modes survey, in which participants reported on potential incentives and barriers to walking and their travel behaviour.
Using machine learning, this study first examined if perceived walkability and demographic characteristics could predict if a person would rarely or frequently walk. Having confirmed the model’s ability to predict behaviour, with a reasonable level of confidence, the study examined which of the numerous potential explanatory variables were the most important in predicting behaviour.
The use of public transport and the qualities of walking as compared to other modes were crucial. The findings suggested developments to the theoretical model linking built environment to walking behaviour (draft Social Model of Walkability), namely including explicitly the broader transport system and the perceptions of walking as compared to and combined with (public transport).
The full paper, co-authored with Professor Erica Hinckson, Associate Professor Melody Smith, and Dr Tom Stewart, is available here. This work was part of my thesis, available here (supervision: Professor Erica Hinckson, Associate Professor Melody Smith, Dr Tom Stewart).