USING A MULTINOMIAL LOGIT MODEL TO STUDY THE MODE CHOICE BEHAVIOR OF COMMUTERS IN FLANDERS

This paper demonstrates the possibility and the viability of combining new data resources with traditional surveys to estimate a mode choice model for work/school trips. These modelling tools are used to evaluate the effects of behaviour change. Its purpose is to examine the travel mode choice preference of commutes to work/school. The proposed model is tested using a travel survey conducted in the Flanders region in Belgium referred to as the Onderzoek Verplaatsingsgedrag 5 (OVG), which translates to research on travel behaviour. The objective is to analyse the correlation between mode choice... Mehr ...

Verfasser: Amit Pothina
Prof. dr. Muhammad Adnan
Dokumenttyp: Artikel
Erscheinungsdatum: 2022
Schlagwörter: Biogeme / Multinomial Logit Model (MNL) / utility function / mode choice model / travel behaviour / travel survey
Sprache: unknown
Permalink: https://search.fid-benelux.de/Record/base-27087753
Datenquelle: BASE; Originalkatalog
Powered By: BASE
Link(s) : https://zenodo.org/record/6645365

This paper demonstrates the possibility and the viability of combining new data resources with traditional surveys to estimate a mode choice model for work/school trips. These modelling tools are used to evaluate the effects of behaviour change. Its purpose is to examine the travel mode choice preference of commutes to work/school. The proposed model is tested using a travel survey conducted in the Flanders region in Belgium referred to as the Onderzoek Verplaatsingsgedrag 5 (OVG), which translates to research on travel behaviour. The objective is to analyse the correlation between mode choice, mode characteristics and socio-economic attributes of individuals. These are computed using a multinomial logit model (MNL) through a utility function. After testing out the significance and correlation of over 50 mode-specific and individual-specific characteristics using the Biogeme package, travel time, education level, and income level were identified as the attributes with the highest utility in the model. The OVG 5 data contained more than 100 individual attributes and mode characteristics, all of which were tested in the model's performance. The final utility function uses the 15 most significant parameters and yields the highest final log-likelihood score compared to the other models. Findings of this research contribute knowledge towards making transport policies in Flanders, for example, inducing a modal shift to more sustainable modes of transport amongst commuters. There are also behavioural implications and transport planning benefits associated with the findings of this research. The dataset was collected until early 2020, thus making it the last pre-COVID travel survey in Flanders. Future research like the OVG 6 and higher can use this model as a base model and build upon it because the study area will remain the same.