关键词: 城市交通, 职住地, 出行活动, 轨迹数据, 聚类算法

Abstract:

The vehicle trajectories contain enrich travel characteristic information. Analyzing the characteristics of travel activities and identifying the location of work, residence, and travel interests from the trajectories could provide the support for the layout plans of transportation facilities and the reduction plans of transportation carbon emission. The research takes the point set that composed of the first starting point and the last ending point of daily travel activities as the possible point set for residence, while the other starting and ending points form the possible point set for workplace and place of interest. On the basis of possible point sets, a method that is based on mean shift clustering and spatiotemporal dual constraints is proposed to determine the location of residence. Combining three conditions: cluster density, average residence time of points within the cluster, and travel time range, the coordinates of residence, workplace, and place of interest are determined. Based on the KD Tree algorithm, adjacent POI data is matched for each type of location coordinate to obtain the specific location and name of the workplace and residence. Based on the identification of workplace and travel interests, the travel activity is characterized by the number of trips, travel distance, and travel time. The stability and difference of travel are used to characterize the regularity of travel. The K-means++ clustering analysis algorithm is used to analyze the characteristic categories of travel activities. Taking 1708 vehicles with travel activities for 34 days in Beijing as an example, the empirical research is conducted based on the driving trajectory data. The research results indicate that the distribution characteristics of workplace and residential areas determined by the proposed method is consistent with practical laws, with high accuracy and reliability. The travel activity feature classification method based on K-means++ algorithm classifies all 1708 travelers into four categories: inactive, active multiple trips, active long-term trips, and active long-distance trips. According to the two indicators of travel regularity (travel stability and work rest travel difference), travelers with changes in travel patterns are classified into four categories: stability active on workdays, stability active on rest days, non-stability active on workdays, and non-stability active on rest days. (Supplement the proportion of each type).

Key words: urban transportation, place of residence, travel activities, trajectory data, clustering algorithm