“…In order to take into account both heterogeneity and spatial factors (e.g., neighbour effect), the EV market model needs to be coupled with a population synthesizer (Pritchard and Miller, 2012), a social network generator (Arentze et al, 2012), and an activity-based travel demand model (Horni et al, 2016;Zhuge et al, 2017). Specifically, population synthesizer is used to generate a synthetic population containing individuals and households, as well as their attributes (e.g., income and car ownership) (Pritchard and Miller, 2012;Zhuge et al, 2016a;, which can be used as the inputs of the MNL models to predict the weights of each factor; the social network generator is used to generate a population-wide social network (Arentze et al, 2012;, so that the three types of social influence can be quantified and the results can be further used as the inputs of the utility function; Activity-based travel demand model, which is used to simulate the daily travel of each individual in the population (Horni et al, M A N U S C R I P T A C C E P T E D ACCEPTED MANUSCRIPT 2016), can be used to quantify the vehicle usage and environmental awareness (e.g., the total amount of vehicular emissions) by aggregating the micro-simulation results. In addition, this paper used four general clusters, namely "Very High", "High", "Medium" and "Low", to group the weights of each factor, using a K-means clustering algorithm with K set to 4.…”