Data Generators for Big Data Analysis

U današnje doba postoje podaci koji su teško dostupni iz više razloga, no problem njihove nedostupnosti, zbog sigurnosti ili privatnosti, može se izbjeći razvojem generatora podataka. U ovom radu testiran je i nadograđen generator podataka za dva skupa koji na dva načina opisuju stvarne podatke prikupljene s društvene mreže Facebook. Oba su skupa dana istom generatoru, čiji su izračuni i rezultati podvrgnuti više analiza, od kvalitete do trajanja izvođenja njihovog stvaranja. Generator uspješno reproducira distribucije i korelacije podataka, te se izračuni odvijaju u prihvatljivom vremenu na računalu slabijih do srednjih performansi. Daljnja nastojanja autora su stvaranje integralnog rješenja za oba dana skupa, a zatim i razvoj generatora detaljnih podataka društvene mreže koja bi odgovarala mreži Facebook na nižoj razini zrnatosti. Nowadays, there is data that is difficult to access for a number of reasons, but its inaccessibility due to security or privacy can be avoided by developing a data generator. In this paper, a data generator for two datasets which describe actual data collected from Facebook is tested and upgraded to suit the task at hand. Both sets were run through the same generator, the results of which were analysed in different ways, from quality to duration of execution. The generator not only successfully reproduces the given distributions and data correlations, but also finishes calculations within an acceptable time period on a computer of a weaker to medium performance. The author's further efforts include creating an integral solution for both given sets, as well as developing a detailed social network data generator which would generate data according to Facebook data of lower granularity.