Exploring Dynamic Clustering for Network Time Series Analysis

This technical abstract discusses a method for exploring dynamic clustering for analysing network time series information. The proposed approach is based on a mixture of records-mining techniques and clustering algorithms geared toward coming across units of time series that percentage the equal conduct. By means of leveraging the marginal distribution of the network facts and the connection between the nodes, this approach allows us to become aware of and exploit temporal patterns throughout a couple of networks. Moreover, a hard and fast method to discover clusters in specific space and temporal scales is also offered. The accuracy of the proposed approach is evaluated using publicly available benchmark datasets. Eventually, a case examination is supplied to demonstrate the effectiveness of the proposed technique. The received consequences display that the proposed approach is suitable for detecting clusters in networked time-collection statistics.