Assessing expected ride-pooling performance with non-deterministic, heterogeneous travellers' behaviour.

Abstract

Introducing social networks to travel demand models could better capture socially induced travel behavior. This paper presents an agent-based approach to forming social networks that match important global characteristics and egocentric homophilies in distance, age, and gender for a population on the order of 106 . Based on data from an egocentric snowball sample, this methodology successfully reproduces homophilies in age and gender, as well as an expected power-law distribution of geographic distance between connections. An initial clique formation heuristic is implemented on top of the homophily calculations. The generated network exhibits preferential attachment between agents of higher degree, in line with more general literature on network formation.