DBSCan clustering algorithms for non-uniform density data and its application in urban rail passenger aggregation distribution

Abstract

With the emergence of all kinds of location services applications, massive location data are collected in real time. A hierarchical fast density clustering algorithm, DBSCAN(density based spatial clustering of applications with noise) algorithm based on Gauss mixture model, is proposed to detect clusters and noises of arbitrary shape in location data. First, the gaussian mixture model is used to fit the probability distribution of the dataset to determine different density levels; then, based on the DBSCAN algorithm, the subdatasets with different density levels are locally clustered, and at the same time, the appropriate seeds are selected to complete the cluster expansion; finally, the subdatasets clustering results are merged. The method validates the clustering effect of the proposed algorithm in terms of clustering accuracy, different noise intensity and time efficiency on the test data of public data sets. The experimental results show that the clustering effect of the proposed algorithm is better than traditional DBSCAN. In addition, the passenger flow data of the night peak period of the actual site is used to identify the uneven distribution of passengers in the station. The result of passenger cluster identification is beneficial to the optimization of service facilities, passenger organization and guidance, abnormal passenger flow evacuation.