DEFINING DYNAMIC SPATIO-TEMPORAL NEIGHBOURHOOD OF NETWORK DATA

To improve the accuracy and efficiency of space-time analysis, spatio-temporal neighbourhoods (STNs) should be investigated and analysed in the classification, prediction and outlier detection of space-time data. So far most researches in space-time analysis use either spatial or temporal neighbourhoods, without considering both time and space at the same time. Moreover, the neighbourhoods are mostly defined intuitively without quantitative measurement. Furthermore, STNs of network data are less investigated compared with other types of data due to the complexity of network structure. This paper investigates the existing approaches of defining STNs and proposes a quantitative method to define STNs of network data in which the topology of the network does not change but the characteristics of the edges (i.e. thematic attribute values) change with time which requires dynamic STNs adapted to the properties of the network. The proposed method is tested by using London traffic network data.