Research And Application Of Railway Co-Travel Travel Group Relationship Prediction Based On Hidden Markov Model

In recent years,with the continuous development of railway transportation network and high-speed railway technology,the speed and comfort of railway travel have been greatly improved,and railway travel has been chosen by more people.With the annual increase in railway passenger trips,the behavior of co-travel has become common in rail trips,such as traveling with family members and traveling with colleagues.The travel behavior of passengers will be affected by their peers,and different travel groups have different travel preferences.For example,when traveling in a family group,the elderly and children in the group will be considered,and they are more inclined to comfort;when traveling in a group of friends composed of young people,they will focus on the sense of experience and freshness.Therefore,the type of co-travel group is the basis for studying the travel preference of the group.Accurate prediction of co-travel group relations will not only help transportation,tourism and other related industries to define the products and services that travel groups interest in,but also provide support for market decision-making in the railway transportation industry.Based on above,the aim of this paper is to develop a methodology for analysing railway passengers’ travelling behavioural using ticket booking data.This study introduces Hidden Markov model to predict passengers group type,i.e.,whether the group of passengers are family members or not.Firstly,the applicability of hidden Markov model in railway data is studied,the railway hidden Markov model is constructed,and the prediction problem is defined based on the model;Secondly,based on the characteristics of railway ticketing data,this paper proposes the quantitative method of co-travel times of railway travelling group;Then,the semi-supervised learning method is used to amplify the labeled samples in a small batch.And the accuracy and consistency of the model prediction are verified by using the actual railway ticket data,the experimental results show that the average accuracy of the model is as high as 0.9573,and the consistency is as high as 0.965 for the same travel group at different times.It can be concluded that the proposed method can effectively and accurately predict the potential peer relationship in railway travel groups.