Aperfeiçoamento da previsão global em séries temporais caóticas

• Autor(a) principal: Alves, Paulo Ricardo Lourenço
• Data de Publicação: 2013
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Biblioteca Digital de Teses e Dissertações da UERJ
• Texto Completo: http://www.bdtd.uerj.br/handle/1/12915
• topic: Time Series; chaos; predictability; Séries temporais; caos; previsibilidade; CNPQ::CIENCIAS EXATAS E DA TERRA::MATEMATICA::MATEMATICA APLICADA::FISICA MATEMATICA

The prediction of future values in temporal series produced by chaotic systems can be applied on several fields of knowledge, such as Astronomy, Economics, Physics, Medicine, Meteorology, and Oceanography. The applied method consists on the reconstruction of the phase space, followed by an improvement term of the forecasting. The routines used for the prediction and analysis of this line of research are part of the TimeS package, which presents encouraging results on their applications. The improvement of the computational routines from the package TimeS is built from the thorough investigation of the minimization applied on obtaining the global map and aims for the enhancement of the accuracy and reduction of computational times. The mathematical basis is established and computational tasks are created. The possibilities of adjust functions are amplified, which can include transcendental terms on the rebuilt vectors components and also possess linear or non-linear terms on the adjustment parameters. The improvement allows more accurate predictions and analysis, which answer important questions regarding prediction methods and improve the possibilities of application on real series.