Aggregating spatio-temporal phenomena at multiple levels of detail

Abstract

Spatio-temporal data are collected at high levels of detail (LoDs). Both spatial and temporal characteristics of data can be expressed at different LoDs. Depending on the phenomenon and the analytical goal, different LoDs can be suitable for a user’s analysis since different LoDs may provide different percep- tions of a phenomenon. It is crucial to model spatio-temporal phenomena having in mind that different LoDs can be useful in their analyses. We propose a granu- larities-based model in order to model spatio-temporal phenomena at multiple LoDs. It defines the concept of LoD and afterwards the atom generalization, gran- ular synthesis and compressed granular syntheses set concepts to express a phe- nomenon at some LoD into a coarser one. This occurs in a semi-automatic way as the user just needs to define functions that create the compressed granular synthe- ses sets. A demonstration case was conducted applied to a real dataset about acci- dents in USA in which the model proposed proved to be useful to reduce the amount and complexity of data when the phenomenon is observed at coarser LoDs than the ones at which data is provided.