SILVA, H. S.; http://lattes.cnpq.br/3672870581352292; SILVA, Hugerles Sales.
Resumo:
This report presents a study on methods for dimensionality reduction, specially the random
projection, based on Johnson-Lindenstrauss lemma, to create sketches of big set of
data, maintaing the same characteristics as the original data. One literature review, beyond
the basic theory of methods to the extraction of attributes in one set of data and the
obtainment of data sketches. Applications in the analyses of data in the area of health care
and meteorology are evidenced and determined parameters like: processing time to ind
certain tendencies, absolute medium error, among others, are analysed and compared.