Resumen de Aplicacions de visualització d'informació georeferenciada
In a context where geolocation tools are becoming more common, statistical data collection and dissemination systems should include new tools to release georeferenced information with precise spatial resolution. At the same time, this accuracy may trigger confidentiality issues for individuals, households or companies to whom this data refers, because it can enable their re-identification. There is a conflict between the dissemination of data at an accurate level of geographical precision and, at the same time, fulfill the different confidentiality regulations on the distribution of statistical data. Statistical offices typically use collection systems based on administrative boundaries to divide the territory and assist data organization. These partitions are also useful to ensure confidentiality but may not be the most appropriate form of dissemination when data should be used to study spatial effects of different phenomena. In this thesis we propose an alternative methodology for data distribution which takes an initial constant size grid, and successively divides the space to obtain an irregular grid that groups the information balancing confidentiality and spatial resolution. Grids produced using this methodology are based on a cell coding system following the indications proposed by the Statistical Office of the European Union, to represent the main characteristics of the population on a unique constant grid. Measuring the importance of spatial effects requires having data at an accurate spatial resolution and, when possible, on a scale close to that of the phenomenon studied. In this regard, this thesis also presents two studies of spatial phenomena conducted with the information available using the current systems based on census tracts. In both studies the investigated phenomena are characterized at a very precise spatial resolution, while the available data to detect possible spatial correlations is aggregated in spatial units with much less resolution. In order to compare the different processes, data must be transformed to the same geographic scales. These transformations may have undesired effects on the results of the analyses. The proposed methodology undertakes the production of more accurate spatial datasets to avoid those possible skewing effects. The methodology is fully developed on a library in the statistical software R, in order to produce data for this environment, but it can also be exported to standard data formats for any other geographic information system. The library is published in the usual R software repository and can be downloaded and installed to publish grid-based datasets from any spatial point data. In the future, this library will allow statistical offices to provide more accurate spatial information to perform spatial analysis.
