Análise de valores extremos para dados de poluição atmosférica na cidade de São Paulo

• Ano de defesa: 2010
• Autor(a) principal: Martins Júnior, Hernani
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: UNIVERSIDADE FEDERAL DE LAVRAS; DEX - Programa de Pós-graduação; UFLA; BRASIL
• Palavras-chave em Português: Poluição atmosférica; Valores extremos; Distribuição generalizada de valores extremos; Valor ao Risco (VaR); Testes de aderência; Air pollution; Extremes values; Generalized extreme value distribution; Significance test; CNPQ_NÃO_INFORMADO
• Link de acesso: http://repositorio.ufla.br/jspui/handle/1/3377

High levels of air pollution have steadily worsened the lives of people in big cities. Therefore it is necessary a rigorous control of these variables. This work tries to model the extreme events of a series of air pollutants in Sao Paulo. For this is the methodology of extreme values applied to stationary time series and independent. The distribution parameters are estimated by Maximum Likelihood Method for the solution of equations likelihood is given by iterative methods of solution. The suitability of the model is tested by the Kolmogorov-Smirnov test. Once you set a template, this, as well as estimates of its parameters are used to calculate the VaR, Value at Risk, which is a risk measure commonly used in the financial market. This application can tell whether the use of VaR is useful in monitoring these atmospheric variables.