Extracting Human Activity Areas from Large-Scale Spatial Data with Varying Densities

Human activity area extraction, a popular research topic, refers to mining meaningful location clusters from raw data. However, varying densities of large-scale spatial data create challenge for existing extraction methods. This proposes novel framework (ELV) aimed at tackling the by using clustering with an adaptive distance parameter and re-segmentation strategy noise recovery. Firstly, was adaptively calculated cluster high-density points, which can reduce uncertainty introduced human subjective factors. Secondly, remaining points were assigned according characteristics clustered more reasonable judgment points. Then, face density problem, designed segment appropriate into low- clusters. Lastly, produced in step recovered unnecessary noise. Compared other algorithms, ELV showed better performance on real-life datasets reached 0.42 Silhouette coefficient (SC) indicator, improvement than 16.67%. ensures reliable results, especially when differences are large, be valuable some applications, such as prediction recommendation.