Adatelemzési folyamat és keretrendszer a közigazgatás számára (Data analysis process and framework for public administrations)

Abstract

The leap forward in artificial intelligence over the last decade, with the continued expansion of the hardware and software platforms that support it, has also taken data analytics to a new level. In principle, this is best understood as a reduction in the need to precisely define processing models, as the tools now available can ensure that by simply providing the raw input data and defining the desired goal, the effective analysis procedure - usually a neural network architecture - is automatically designed through a machine learning process. As this trend is expected to increase in the future, it is advisable to build the analysis procedures to fit into this framework. Accordingly, emphasis should be put on pre-processing the datasets to be processed, potentially from different domains, so that the complete data set can be passed to the analytical architecture. Since the interpretability of the analysis results must be made amenable to human use, visualisation techniques may typically be used for this purpose. It makes sense to integrate the visualisation technique into the overall analysis framework for reasons of efficiency, i.e. the visualisation tool is built directly onto the output of the analysis architecture and its internal data representation, if, for example, the presentation of the relationships between the input data is useful for justifying the decision.

Keywords

• data analysis
• framework
• data analysis plan