The mixed logit model, and so on. The general principle of the discrete option model is stochastic utility theory; that is certainly, when decision maker n faces the option, there are i decision schemes, and also the preference for any specific choice scheme i is usually described by the utility value Unit of your selected object. Vnit would be the observable component from the utility function, also called the fixed utility function, and nit may be the random error element of your utility function. The distribution form with the random error function nit determines 20(S)-Hydroxycholesterol Activator distinctive discrete choice models. Therefore, the utility function of exit i selected by passenger n in subway emergency evacuation scenario t is often characterized as Formula (1): Unit = Vnit nit three.two.2. The Observable Part on the Utility Function of Logit The independent variables of your utility function are “Dist”, “Pedestrian flow” and “Crowd density” The observable component with the passenger utility function can be expressed by Formula (two): Vnit = 1n ( Dist)nit 2n (Crowd density)nit 3n ( Pedestrain f low)nit (two) (1)where ( Dist)nit would be the distance in the passenger n to exit i in experimental scenario t, (Crowd density)nit would be the number of passengers at exit i in experimental scenario t, ( Pedestrain f low)nit will be the variety of passengers flowing to exit i seen by the passenger n in experimental scenario t and 1n , 2n and 3n would be the parameter coefficients. three.two.3. The Random Parameter Logit Model The principle purpose of this paper is to study the heterogeneity of passenger evacuation preference. On the other hand, some logit models can’t determine the heterogeneity of preferences, for instance the conditional logit model, the nested logit model, and so on., since these logit models commonly make use of the maximum likelihood technique for parameter estimation, however the maximum likelihood estimation approach assumes that the probability of event occurrence is only determined by the variables in the model, which ignore the influence of factors outdoors the model and uncertain elements around the probability of event occurrence. These logit models above did not look at the limitations of individual variations as well as the IIA hypothesis (the IIA hypothesis states that for any person, the ratio with the probability of selecting two options is independent of the presence of attributes of any other option) [28], so the random parameter logit model is proposed to resolve this problem. The random parameter logit model sets the coefficient as random, which can far better capture the heterogeneity amongst decision makers. The experimental information come from the selection outcomes of participants for distinctive evacuation scenarios. The variations of those things may result in heterogeneity. The random parameter logit model has been proven to become a very good indicator of this heterogeneity [31]. The random error term nit of your random parameter logit model follows Gumble distribution, as shown in Formula (three): f ( nit ) = e- nit e-nit(three)Sustainability 2021, 13,6 ofThe utility coefficient n follows regular distribution, and n might be expressed by Formula (4): n = n (4) The random parameter logit model could be expressed by Formula (five): Pnit =nTeVnitn j =1 et =IVnjt(n )dn(five)exactly where could be the average in the coefficients, n may be the vector of independent normal variables, could be the Cholesky element of the covariance PK 11195 web matrix and (n ) could be the probability density function. three.3. Calculation Process of Character Traits Costa [32] proposed a 5 personality traits model, such as 5 traits called O.