Quantitative Determination Method for Traffic Analysis Zone Generation and Attraction Points

With the goal to improve the accuracy of traffic assignment without increasing the additional workload of data collection and data processing, a quantitative method to determine traffic analysis zone generation and attraction points is proposed based on clustering analysis and land use data, including area of interest, the number of households in residential community, and building storey data. Taking the morning rush hour as traffic analysis time period, multi-source urban land use data were collected using web data crawling technology. To obtain traffic generation and attraction points with traffic activity intensity weights, clustering analysis was conducted on residential and non-residential land which could reflect the regularity of traffic demand distribution in traffic analysis zone. Then, automatic subdivision of traffic analysis zone was realized by distributing traffic demand to its generation and attraction points based on traffic activity intensity weights. The results from the practice show that there exists obvious regional aggregation and the clustering results of residential and non-residential land are generally inconsistent; by replacing the original 332 centroids of traffic analysis zone with 665 traffic generation points and 606 traffic attraction points determined by this method, the proportion of intra-zone motor trips, which are not assigned to the road network, decreases from 3.55% to zero, and the relative error of sampling sections’ forecasting results decrease by 15.99% and 9.36% in peak value and average value respectively. It is shown that traffic generation and attraction points quantitative method can improve the accuracy of traffic assignment while maintaining the existing traffic analysis zone division.