Track Mining Based on Density Clustering and Fuzzy C-Means

The spatial-temporal features extracted from trajectory data inherently contain rich knowledge of geography and application semantics. Trajectory data mining can obtain implicit, unknown and potentially useful information and apply it to areas such as urban management, traffic planning, traffic navigation, meteorological detection, and business decision-making. This paper combines density-based clustering methods and fuzzy C-means methods to mine urban hotspots. Among them, the DBSCAN algorithm is exploited for data preprocessing, leading to a significant improvement when filtering the noise data and outliers. The fuzzy C-means algorithm divides the urban area into several sub-regions automatically, supplemented by low-speed informationThe experiment results show that the proposed method can obviously excavate the urban hotspots.