A Study On Heterogeneity Of Commuters' Choice Of Transport Based On Latent Class Models

Commuting is the main travel mode of city residents,and is also the most important one;it has a decisive impact to a city's daily traffic conditions,The most significant part of urban traffic congestion is from commuting.Limited to urban transportation resources,increasing the supply of roads and improving the quality of transportation services will not be a permanent measure.In recent years,a new idea in the field of transportation behavior research based on the personal characteristics and preferences of the individual appears.Discrete choice model is a classic way to study the travelers' choice of path and mode,it is developing with the improving needs of people,the mixed logit model can predict the selection method according to individual random preference factors.While the latent class model is a further study of the immeasurable factors,and it can control variables to a great extent,excluding mutual interference factors,with potential variables to measure the relationship between the explicit variables to be investigated.New classifications are also made at the same time;the proportion of the crowds can be explored as well,which is conducive to our further research and guidance when targeted.Latent class model is an important complement to the choice of heterogeneity in commuting behavior.The main contents and conclusions of this paper are as follows:First of all,to understand the spatial characteristics,strength characteristics,purpose characteristics,we go through literatures of urban commuter travel.Secondly,we combed the existing methods and conclusions systematically,based on the traditional model of travel behavior selection methods,recognizing that in many cases,enriching roads can only have a significant effect in the short term.Then,the mathematical principles of latent class models are introduced in detail,it is necessary to use several basic examples to illustrate its mathematical principles and parameter estimation methods.Finally,the data were collected through the stated preference survey method,latent class model was used to classify the survey population,and each category was analyzed and named.In this paper,SP survey is conducted through economic level,weather conditions,commuting distance,commuting cost,attendance,comfort,identity,each of the attribute selection is divided into three levels.The mathematical principles of latent class model is introduced and through LatentGold software,according to the data distribution and structure characteristic,the subjects were divided into three categories,potential probability of potential categories are respectively 0.5487,0.3914,0.0599,they are named:"Cost-sensitive commuters","Taking-work-seriously commuters","Focusing-on-experience commuters",and each conditional probability is estimated so the selection heterogeneity of each category can be analyzed.