HMHelia Mohammadi-Mavi

Helia Mohammadi-Mavi

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Helia Mohammadi-Mavi has a total of 9 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    From Personas to Travel Diaries: Evaluating LLM Agents for Travel Behavior and Mode Choice Simulation

    by Sorena Vosoughkhosravi, Abdolmajid Erfani, Helia Mohammadi-Mavi

    Accurate simulation of individual travel behavior is central to transportation planning, yet conventional travel-demand models often require extensive behavioral rules, calibration, and detailed mobility data. This study evaluates demographically conditioned large language model (LLM) agents as an emerging approach for generating daily travel diaries from limited person-level information. Using 2024 American Time Use Survey (ATUS) microdata, 150 employed respondents are converted into traveler personas defined by eight sociodemographic attributes. Four LLMs, including GPT-5.6, Claude Opus 4.8, Llama 3.1 8B, and Phi-4-mini generate complete 24-hour travel diaries across three runs. A three-tier validation framework assesses trip generation, mode choice, and departure timing, supplemented by tests of telework behavior, reproducibility, output validity, and cost. The models generally reproduced major trip purposes and, except Phi-4-mini, closely captured the strong private-vehicle dominance observed in ATUS. However, all models over-generated travel, rarely reproduced no-travel days, and concentrated departures at stylized clock times. Frontier models were substantially more reproducible, while Llama 3.1 8B achieved near-frontier mode-choice fidelity at zero marginal inference cost. The findings demonstrate both the potential and current limitations of demographically conditioned LLM agents for travel-behavior simulation.

  • Open Access

    Should We Blame COVID-19 for the Decline in Transit Ridership, or Was It Merely a Stimulant?

    by Helia Mohammadi-Mavi, Mustafa Fardin, Andisheh Ranjbari

    Public transit ridership significantly declined during the COVID-19 pandemic and has not yet rebounded, even though the virus substantially subsided, and people have mostly returned to their normal lives. This suggests that there are reasons for this drop that go beyond health concerns and work from home arrangements. Focusing on pre-pandemic transit users, this study aims at understanding the most influential reasons for the transit ridership decline by digging into people's attitudes and perceptions. Exploiting a dataset of survey responses from the Puget Sound region in the U.S. Pacific Northwest, we performed two analyses: Structural Equation Modeling, and user clustering (based on attitudes) combined with statistical tests. The core of both analyses was people's attitudes/perceptions toward three matters: sharing rides with strangers, doing activities online vs. in person, and pandemic health risks. The results showed while those three all play a role in people's post-pandemic reduced transit use frequency, the reluctance to share rides with others is the only significant factor and its magnitude is three times larger than the health concerns/fears associated with the pandemic. The latter was found to have rather an indirect effect on reducing transit ridership, through its significant and direct impact on both ridesharing reluctance and online activities preference. In other words, the pandemic created/stimulated a desire in people to do certain activities (or a portion of those) online and intensified the reluctance to share rides with strangers in them, and these two attitudes stayed with people post pandemic, resulting in yet low transit ridership.

  • Open Access

    COVID-19 Impacts on Online and In-Store Shopping Behaviors: Why they Happened and Whether they Will Last Post Pandemic

    by Jorge Manuel Diaz-Gutierrez, Helia Mohammadi-Mavi, Andisheh Ranjbari

    Throughout the COVID-19 pandemic, people’s online and in-store shopping behaviors changed significantly. As the pandemic subsides, key questions are why those changes happened, whether they are expected to stay, and, if so, to what extent. We answered those questions by analyzing a quasi-longitudinal survey dataset of the Puget Sound residents (Washington, U.S.). We deployed structural equation modeling (SEM) to build separate models for food, grocery, and other items shopping to explore the factors affecting such changes. The results revealed that people’s online and in-store shopping frequencies during the pandemic were affected by their perceived health risk, attitudes toward shopping, and pre-pandemic shopping frequencies. Similarly, it was shown that how frequently people expect to shop post pandemic is influenced by their attitudes toward shopping, changes during the pandemic, and their pre-pandemic frequencies. We also classified respondents into five groups, based on their current and expected future shopping behavior changes, and performed a descriptive analysis. The five groups— Increasers , Decreasers , Steady Users , Returnees , and Future Changers —exhibited different trends across online and in-store activities for shopping different goods. The analysis results showed that, while 25% of the respondents increased their online shopping, only 8% to 13% decreased their in-store activities, implying that online shopping did not completely substitute in-store shopping. Moreover, we found that online shopping is a substitution for in-store shopping for groceries, while it complements in-store shopping for food and other items. Additionally, more than 75% of new online shoppers expect to keep purchasing online, while 63%–85% of in-store Decreasers plan to return to their pre-pandemic frequencies.