Study On The Travel Behavior Based On GPS Data

For residents of investigation and analysis of resident trip characteristics and master the resident trip distribution is an important part of scientific planning and effective traffic management system.The traditional questionnaire survey method is very vulnerable to the influence of the subjective consciousness of the respondents.The method of obtaining resident travel data by GPS geographic location and time information can avoid this shortcoming,which provides comprehensive and reliable data support for traffic modeling and the construction of traffic operation monitoring platform.Therefore,the use of GPS devices to obtain travel trajectory data to study residents' travel behavior and grasp the characteristics of traffic distribution has become a hot topic in recent years.Among them,through mining and analyzing GPS trajectory data,obtaining complete travel chain information and identifying residents' travel modes and purposes are the main contents of the study on travel behaviors.In this paper,based on the GPS trajectory data of travelers,machine learning algorithms such as C4.5 decision tree,BP neural network and random forest are adopted to identify the travel mode and the purpose of travel based on geographic location information.First of all the collected data by excluding certain rules of pretreatment,extraction from travel characteristics.In order to improve the quality of data,the velocity and acceleration of the trajectory data are smoothed by five times smoothing.Secondly,in order to determine the traveler's residence information from the discrete trajectory data,a spatial-temporal clustering algorithm for three-dimensional data processing is proposed by using the set time and distance parameters.According to residents' stay location and time information,the travel segments in the trajectory data were extracted,and the characteristic parameters of each travel segment were extracted.Three machine learning algorithms,namely C4.5 decision tree,BP neural network and random forest,were used to identify the traffic travel modes.The results showed that the random forest algorithm had the best effect and the highest accuracy rate was up to 90%.Finally,in order to make up for the limitation of judging travel purpose based on rules in the past,POI information around the traveler's stay area was extracted and information entropy was calculated by virtue of virtue map service,and the travel purpose of residents was identified by combining the characteristics of residence duration and residence start time obtained by the spatio-temporal clustering algorithm.The results show that the research on travel behavior based on GPS data can be well applied to the identification of travel modes and the analysis of travel characteristics,which plays an important role in the promotion of GPS trajectory data in the research field of residents' travel,the analysis of residents' travel behavior rules and the promotion of the healthy development of urban traffic system.There are two major innovations in this paper:first,in order to divide the trajectory data into the travel segment of a single mode of transportation,a three-dimensional data spatio-temporal clustering algorithm is designed to identify the location and time of the traveler's stay.Second,in recognition of traffic transportation,the use of the people use fewer random forest algorithm,it is concluded that the precision of the than C4.5 decision tree,BP neural network algorithm,such as higher recognition accuracy,and in the category of travel way of recognition,joined the often ignore the recognition of the electric bicycle,the recognition accuracy as high as 88%.