Abstract
Previous studies often overlook nonlinear relationships between built environment characteristics and travel satisfaction, and few examine the interaction effects of these characteristics on travel satisfaction. Using gradient boosting decision trees on a dataset of 1167 respondents collected from Harbin, China, in 2021, we estimated the nonlinear and interaction effects of built environment characteristics on travel satisfaction. We found that perceived built environment attributes, such as infrastructure and safety for walking, aesthetics, physical barriers, and land-use mix, are key predictors of residents' travel satisfaction, and their relationships are mostly nonlinear. Furthermore, built environment characteristics exhibit three patterns of interaction effects on travel satisfaction: compensation, suppression, and reinforcement. These findings can help planners better assess the costs and benefits of environmental improvement plans.
| Original language | English (US) |
|---|---|
| Article number | 104111 |
| Journal | Journal of Transport Geography |
| Volume | 123 |
| DOIs | |
| State | Published - Feb 2025 |
Bibliographical note
Publisher Copyright:© 2025 Elsevier Ltd
Keywords
- Built environment
- Interaction effect
- Machine learning
- Nonlinear effect
- Three-factor theory
- Travel satisfaction