Modelling Social Preferences of Ride-sharing
Makhija, Ramandeep Singh Manjeet Singh
MetadataShow full item record
Ride-sharing also known as carpooling is sharing of journeys from one place to another so that more than one person can travel together and obviate others from driving to the locations themselves. By having more than one people sharing a vehicle, ride sharing avails to reduce personal expenses such as fuel, driving stress and tolls. The objective of this paper is to model the social preferences of people ride-sharing together i.e. to identify challenges and barriers people face to adopt ride-sharing and model their social preferences to obtain the final prediction for probabilities. First from the literature problems and challenges were identified which people face to adopt ride-sharing. Then a questionnaire was created as a survey which was conducted among the citizens of the United States to know the preferences and attributes of the ride-sharers. Further, using the data collected from the survey, a discrete mode choice analysis was performed. There were 13 main latent variables used for modeling. These 13 variables were grouped into four factors using factor analysis. Finally, four models were created using ordinal logistic regression to predict the probabilities of a person enjoying his/her shared rides in social aspects from his attributes and preferences. After the estimation of 4 final models we estimated a matching index using preference probabilities from the models that give a compatibility ratio between riders.Keywords: ride-sharing behavior; survey; social preferences; ordinal logistic regression; factor analysis; latent variables.