A Fuzzy Clustering Based Method for the Spatio-Temporal Data Analysis

Abstract

Spatiotemporal data is a type of data that is collected by the sensors. This type of data has two spatial and temporal dimensions. There are many challenges in analyzing spatiotemporal big data. Common evaluation metrics of clustering methods are not appropriate for spatiotemporal data. Previous clustering methods and the conventional evaluation metrics are efficient for data like time series with only one segment. Therefore, other metrics are required to evaluate the clustering of such data. In this study, energy function, reconstruction, and prediction metrics are used to evaluate the quality of spatiotemporal data clustering. The purpose of this study is to minimize the energy function using the Fuzzy C-Mean method on spatiotemporal data. The obtained results are compared with those obtained using k-medoid, DBSCAN, COBWEB, X-means, and TLBO. Also, the energy function, reconstruction, and prediction metrics are used to evaluate the quality of the clusters. The clustering methods are implemented on the dataset of parking located in the CBD area of Australia.