Bayesian Hierarchical Spatio-temporal Data Analysis Toolbox for GIS

Scientific Systems proposes to develop a spatio-temporal data analysis toolbox to seamlessly integrate into GIS software so that users can invoke the toolbox for advanced statistical analyses without explicitly leaving their GIS software. The toolbox will have the capabilities for users to interpolate, infer, predict and interpret spatio-temporal data. The toolbox will be capable of handling issues typically encountered in real-life applications such as missing data, misaligned data and mixed data types. Furthermore, the toolbox will be built with the ability to leverage other packages and to ensure future extensibility. Our goal in Phase 1 is to identify the GIS software and the spatio-temporal analysis techniques that are best suited for the spatio-temporal data of customers"interest. The evaluation and selection of the spatio-temporal analysis techniques and the GIS software will lay the foundation for the integration of the spatio-temporal software tools into the GIS software of choice in Phase 2. Our approach is based on Bayesian hierarchical modeling, which permits very general models for temporally and geographically referenced data. Under the Bayesian hierarchical framework, we also incorporate an innovative approach that can adapt the model to sudden changes in data property.