Descoberta automatizada de associações com o uso de algoritmo Apriori como técnica de mineração de dados

• Autor(a) principal: ALMEIDA, Derciley Cunha de
• Data de Publicação: 2011
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional da UFG
• Texto Completo: http://repositorio.bc.ufg.br/tede/handle/tde/966
• topic: 1. Mineração de dados; 2. Algoritmo Apriori; 3.Descoberta de associações; Mineração de dados; Descoberta de associações; Apriori; Banco de dados; WEKA; Data mining; Association discovery; Apriori; Databases; WEKA; CNPQ::ENGENHARIAS

Nowadays, the use of modern information systems allows the storage and management of increasingly large amounts of data. On the other hand, the full analysis and the maximum extraction of useful information from this universe of available data present considerable challenges in view of inherent human limitations. This dissertation deals with the subject of data mining, which is the use of technology resources in order to extract information from databases in an automated way. One of the possibilities offered by data mining technologies is the automated search for possible associations within data. Information about such associations can be useful for understanding cause and effect relationships between the involved variables in data analysis for decision making. There are several data mining techniques and many of them can be used for discovering associations. The main goal of this work is to study a particular method for automated search of associations called Apriori , evaluating its capabilities and outcomes. The study focuses on the problem of improving the Apriori algorithm results, taking into consideration that the results of the data mining process might be improved if the data are prepared specifically for Apriori application. The conclusions are drawn from a case study in which the Apriori algorithm was applied to a database with information on drug distribution at a health institute. The results of two experiments are considered in order to evaluate the influence of data preprocessing on the Apriori algorithm's performance. It was found that the Apriori algorithm yields satisfactory results on the discovery of association in data; however, for best results, it is advisable that the data be prepared in advance, specifically for the Apriori application, otherwise many associations in the database might be left undiscovered.