Heterogeneous Activity Generation for a Synthetic Population: Synthetic Sweden Mobility (SySMo) Model

Abstract

The heterogeneous diversity of activity patterns within population groups has often been disregarded in most literature, resulting an unrealistic representation of travel behavior. We develop a stochastic approach combined with machine learning (ML) to generate heterogeneous activity in a synthetic population. We implement the novel methodology to model the mobility pattern of the synthetic population of Sweden. Comparing the simulated activity schedules with survey data, our results show that the methodology generates realistic mobility pattern of individuals. The proposed model with a realistic representation of activities can make an unique contribution to capture the complexity of travel behaviors, thus better inform policies to improve future transportation systems.