Putrenko V. Methodology for the intellectual analysis of geospatial data for sustainable development goals

The dissertation is devoted to the development of the methodology of the intellectual analysis of geospatial data for use in decision-making systems in the research tasks of sustainable development of the regions of Ukraine in the context of improving the quality and safety of people's lives. Support for management decisions in the management of geographically distributed systems is based on the use of geospatial information, which in turn requires the use of techniques and methods for intelligent analysis and processing of geospatial data to achieve the goals of sustainable development. The main reasons for this are the development of the geospatial data industry, the transition to management based on the use of geoinformation systems, the creation of new systems and methods for collecting geospatial data, including geocoding and geotagging of information, the accumulation of large volumes of geospatial data, which require the latest methods of analysis and the search for the laws in their structure. Therefore, intelligent data analysis opens up new opportunities for finding optimal managerial decisions at all levels of territorial management. In dissertation work the important scientific-applied problem of modeling of parameters of sustainable development with the use of geospatial data on the basis of system approach is solved. The concept of classification has been developed and the expediency of using intelligent methods of analysis of geospatial data for scenario modeling of sustainable development on the basis of methods of system analysis is substantiated. The theoretical and methodological approaches to the formalization of concepts and models of representation of geospatial data are developed on the basis of the paradigm of discrete and continual features of the space of three-dimensional space and its time-shift. In order to conceptualize relations in the geographic space, the concept of geoinformation space as a set of information coordinated computer models of the investigated geospatial is introduced. The primary purpose of using Geospatial Data Intelligence (GDI) is to search for patterns and relationships in large datasets that contain spatially coordinated binding. Therefore, the use of GDI as part of the DSS in territorial management and forecasting is an important and relevant tool for substantiating management decisions. The implementation process of GDI is stable and iterative in order to find optimal analysis results. The main directions of GDI are classification, clustering, rules of associations, geostatistics and geo-visualization, which together form the methodology of systematic intellectual analysis to support decision-making. GDI improves data processing efficiency with other data analysis methods based on different information platforms. Basic methods of geospatial data mining are determined by the type of data distribution and the hypothesis of the probability of estimating the occurrence of anomalous values over a limited spatial distance. The spatial autocorrelation between geospatial objects (Global I Moran Index), the mean nearest neighbor index, the Moran Local Index is used to determine these parameters. For the purposes of spatial clustering, hotspot analysis (Getis-Ord Gi) is used, and the grouping of objects is measured using the Kalinski-Kharabaz pseudo-F statistics. Spatial clustering is determined by the spatial constraints of the topology. An important component of the intellectual analysis of geospatial data is the modeling of spatial relationships by back-distance methods, ranges of distances, zones of indifference, adjacency and neighborhood. An approach to the analysis of big geospatial data by their two-level analysis with the help of data organization in space-time cubes, where based on methods of spatial clustering, the allocation of information patterns of data is developed.