Clustering analysis of manipulate errors for railway transport based on DBSCAN

Aiming at the problems existed in the process of mining manipulate errors for railway transport, an improved DBSCAN algorithm is presented. This method according to the monitoring data of operation of the train, mining railway transportation driver manipulate errors. Algorithm improved DBSCAN in dealing with boundary shared object of cluster by average distance function of shared object. As a result it efficaciously avoids over-segmentation and produces better clustering results. Experimental results show that this method is effective.