Hybrid random utility-random regret model in the presence of preference heterogeneity, modeling drivers’ actions