MiMi : plataforma computacional para mineração de dados micrometeorológicos

• Ano de defesa: 2015
• Autor(a) principal: Oliveira, Allan Gonçalves de
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Mato Grosso; Brasil; Instituto de Física (IF); UFMT CUC - Cuiabá; Programa de Pós-Graduação em Física Ambiental
• Palavras-chave em Português: Mineração de dados; Séries temporais; Micrometeorologia; CNPQ::CIENCIAS EXATAS E DA TERRA::FISICA; Data mining; Time series; Micrometeorological
• Link de acesso: http://ri.ufmt.br/handle/1/2178

The evolution of data acquisition capabilities in electronics devices in recent years created new challenges to the management and analysis of data collected. Researches involving micrometeorological data have been benefited from these developments and also experienced difficulties in relation to data analysis. Thus, the use of computational techniques to aid the process of data analysis is mandatory for their users. This work presents the development of MiMi Platform, a computational platform for mining micrometeorological data. The MiMi platform is intended to provide a general and flexible component based software architecture, which main characteristic involves an algebraic approach for data mining, where the time series are treated as embedded data type in the platform and algorithms that handle them are operators. With the definition of operand and operators, is possible to define domain expressions that represent a specific processing flow for each data mining activity in a time series domain. For this, we defined a domain manager that performs multiple operators in a predefined order. The MiMi platform was validated with the execution of three data mining activities, clustering, similarity search and unknown patterns detection, the tests were performed using two real dataset from micrometeorological stations localized in Mato Grosso, Brazil.