Research On Market Segmentation Of Taxi-hailing Based On Latent Class Model

The Taxi-hailing is favored by many residents due to its lower price,clearly waiting time and cashless payment in the context of "Internet +".However,with the standardization of the market and the diversification of users' demand,the competition core of the Taxi-hailing enterprises is no longer focused on the crazy “subsidy”,but to provide differentiated and diversified services to achieve sustainable development.Therefore,the writer classifies the Taxi-hailing users from the perspective of travel behavior to properly cater to demand and rationally optimize the supply of the limited transportation resources,which are of great practical significance for the Taxi-hailing enterprises.Based on the background of urban comprehensive transportation,the writer chose six travel modes including rail,bus,taxi,express,premier,and car,as alternatives,and some influencing factors such as individual socio-economic attributes,travel-related attributes,and specific attributes of the Taxi-hailing as the attributes to establish a latent class model.Among these,the writer took the “car ownership” as a covariate to conduct the D-efficient experimental design,generating 12 different treatment combinations for the car owners and non-owners.The individual socio-economic characteristics and the Taxi-hailing using characteristics were taken as the revealed preference survey items.A stratified random sampling survey based on age and gender was conducted in Chengdu.Through data screening and effectiveness check,a total of 520 valid questionnaires were collected.Then,according to the results of the questionnaires,preliminary statistical analysis was carried out on the characteristics of Taxi-hailing travel in Chengdu.Based on the survey data,the Taxi-hailing choice behavior model with effect coding was carried out.The results show that latent class model significantly improves the fit compared to the multinomial logit model,and income,time,cost,Taxi-hailing dependence,car ownership,etc.are important basis for classifying Taxi-hailing users.According to the significant factors,the Taxi-hailing users can be divided into four classes,named “Relying on Taxi-hailing type”,“Focus on travel services type”,“Time-cost sensitive type” and “Rarely using Taxi-hailing type”.The class probabilities are 30.6%,15.3%,23.8%,and 30.2%,respectively.Class 1 people have a strong dependence on the Taxi-hailing platform,and increasing the quantities and response speed of Taxi-hailing will improve their stickiness.Class 2 people are willing to consume for better travel quality.Shortening the response time and improving travel service of the Taxi-hailing is conducive to attracting them.Class 3 people are more sensitive to travel time and cost,but they are willing to choose to travel by Taxihailing when traveling for a long-distance.So,discounts such as long-distance coupons are attractive to such users.Class 4 people use less of the Taxi-hailing,who are very concerned about the purpose of travel,the Taxi-hailing only shows a certain advantage in the purpose of going to airport or high-speed rail station,indicating that the Taxi-hailing companies can target the airport or high-speed rail station develop a special connection service to attract such users,among which older users are more likely to turn to the Taxi-hailing to travel.Finally,the elastic and marginal effects analysis of the key variables in the model were carried out.The results show that selection probability of the premier is full of flexibility to the cost,while the express is inelastic,which shows that for the Taxi-hailing platform,the price adjustment of the express has little impact on the revenue management,but the premier is the opposite.