RSAPP, um algoritmo baseado em rough sets para auxílio ao processo de descoberta de conhecimento em banco de dados

Detalhes bibliográficos
Ano de defesa: 2013
Autor(a) principal: Silveira, Juliano Gomes da lattes
Orientador(a): Ruiz, Duncan Dubugras Alcoba lattes
Banca de defesa: Não Informado pela instituição
Tipo de documento: Dissertação
Tipo de acesso: Acesso aberto
Idioma: por
Instituição de defesa: Pontifícia Universidade Católica do Rio Grande do Sul
Programa de Pós-Graduação: Programa de Pós-Graduação em Ciência da Computação
Departamento: Faculdade de Informáca
País: BR
Palavras-chave em Português:
Área do conhecimento CNPq:
Link de acesso: http://tede2.pucrs.br/tede2/handle/tede/5237
Resumo: Techniques of Business Intelligence (BI) became one of the main allies of organizations in tasks of transforming data into knowledge, supporting the middle and upper management levels in decision making. BI tools in their composition are based on techniques of knowledge management, such as Data Warehouse (DW), OLAP (Online Analytical Processing), Data Mining (DM), among others. In this context, it is observed that in many case, DM projects become unfeasible by some factors, such as project costs, duration and specially the uncertainty in obtaining results that return the investment spent on the project. This work seeks to minimize these factors through a diagnosis on data, by an algorithm based on Rough Sets Theory. The algorithm, named Rough Set App (RSAPP) aims to create a diagnosis on data persisted in DW, in order to map which attributes have the greatest potential for generating more accurate mining models and more interesting results. Thus, it is expected that the diagnosis generated by RSAPP can complement the KDD (Knowledge Discovery in Database) process, reducing the time spent on activities of understanding and reducing data dimensionality. This work presents a detailed description about the implemented algorithm, as well as the report of the executed tests. At the end there is an empirical analysis of the results, in order to estimate the effectiveness of the proposed algorithm.