A Case Study of Stream and Density-Based Clustering for Floating Car Data

Abstract

Extracting stay points from location data is important for the detection of point-of-interests or facilities. In this paper, we develop a method to extract stay points via density-based clustering algorithms, by using positional data gathered from floating cars (FCD). Through our analysis, we found a characteristic of FCD when extracting stay-points: the extracted clusters are highly affected by the space-time range of input data. We discuss this characteristic and future directions by showing the clustering results from real data.