A text mining approach to elicit public perception of bike-sharing systems

Bike-sharing has globally emerged as an alternative travel mode. Previous studies have used either ridership data or structured questionnaires to understand public perception. This study contributes with the first application of a text network approach, analyzing open-ended text responses from over 700 Seattle residents regarding their perceptions of docked and dockless bike-sharing systems. The text network approach creates a network based on keyword frequency and co-occurrence. The analysis shows Seattle residents appreciate dockless systems for flexibility but dislike blocked sidewalks and low helmet usage. Negative perceptions of docked systems were varied, indicating improvement challenges. The method explores heterogeneity in user perceptions based on experience, and findings are consistent with econometric approaches, suggesting a hybrid approach for reliable policy recommendations.

Boniphace Kutela, Neema Langa, Sia Mwende, Emmanuel Kidando, Angela E. Kitali, Prateek Bansal