| Popis: |
U ovom radu je predstavljena uporaba Bayesovih mreža na problematici širenja pandemije COVID-19 virusa. Rad objedinjuje dvije analize. Prva analiza se dotiče analize nevremenskih podataka Bayesovim mrežama. Druga analiza obrađuje vremenske podatke novozaraženih, novotestiranih i novoumrlih dinamičkim Bayesovim mrežama. Također, analiza prikazuje utjecaj određenih demografskih, geografskih i ekonomskih faktora na širenje COVID-19 virusa. Kao podloga analize korištene su strukture Bayesovih mreža naučene algoritmima učenja strukture. Također, rad prikazuje prikladnu pripremu podataka za obradu Bayesovim mrežama. This thesis presents the usage of Bayesian networks as a tool for pandemic data analysis related to COVID-19 pandemic. The thesis is split into two parts. First part analyses static data using Bayesian networks. The second part displays methods for analysing dynamic (time series) data of new cases, tests and deaths using dynamic Bayesian networks. Furthermore, the analysis shows causal effects of certain demographic, geographic and economic factors on the spread of COVID-19 virus. The basis of the analysis is presented by the usage of structure learning algorithms for Bayesian networks. Also, the thesis presents complementary data preparation techniques used for Bayesian network analysis. |
