Understanding the expectations of parents regarding their children's school commuting by public transport using latent Dirichlet Allocation.

• Before-and-after impact analysis of public transport (PT) interventions is crucial. • Open-ended (OE) questions complement the information of closed-ended (CE) questions. • OE questions generate insights to adjust CE questions in follow-up surveys. • Latent Dirichlet Allocation (LDA) is reliable for capturing PT users' perceptions. • LDA converts OE replies into useful insights for follow-up marketing interventions. Parents' perceptions regarding public transport and active modes influence the youth's acceptance and support for sustainable school commuting. Urban mobility surveys can gather such insights by utilizing closed and open-ended questions. The latter, particularly, holds the potential for nuanced expectations and insights from Public Transport (PT) users, often absent in closed-ended responses. This paper proposes a methodology utilizing Latent Dirichlet Allocation (LDA) to extract valuable information from open-ended survey responses, shedding light on parents' expectations regarding their children's school commute via PT. Analyzing responses from two surveys involving 448 households, with a focus on parents in the Lisbon Metro Area, spanning the school years of 2017–2018 and 2018–2019, and pre-and post-field interventions, our study employs LDA to assess households' criticisms and recommendations for improving public transport services. Our findings illustrate a shift from general criticisms in the initial survey to proactive suggestions in the subsequent one, aligning with marketing efforts to foster more sustainable school commuting with PT. Empirically, our study underscores LDA's efficacy in capturing users' feedback often neglected by closed-ended questions. Effective preprocessing of textual data facilitates streamlined field interventions. Overall, our contribution provides user-centered insights to inform PT policymakers, promoting the incorporation of user-driven enhancements. [ABSTRACT FROM AUTHOR]