Mode Choice Latent Class Estimation on Mobile Network Data

• In this paper we use a nested latent class logit specification to define and estimate a large-scale mode choice demand forecasting model. We estimate this model based on mobile phone network data translated to roughly 100 000 long-distance trips within Sweden, achieving convergence of the model and credible parameter estimates. We develop methods to address two problems stemming from the nature of this data: the difficulties of distinguishing bus trips from car trips (since they share the same infrastructure) and distinguishing business from private trips (since trip purpose is unknown). To address the first issue, we estimate a nested logit model with an artificial nest that accounts for the differences in utility between bus and car. To address the latter issue, we estimate a latent class model, identifying classes of trips interpreted as private and business trips. Addressing these two issues substantially improves model fit. 

Ämnesord

TEKNIK OCH TEKNOLOGIER  -- Samhällsbyggnadsteknik -- Transportteknik och logistik (hsv//swe)

ENGINEERING AND TECHNOLOGY  -- Civil Engineering -- Transport Systems and Logistics (hsv//eng)

Publikations- och innehållstyp

vet (ämneskategori)

ovr (ämneskategori)