Análise de crédito bancário com o uso de DATA MINING : redes neurais e árvores de decisao
Resumo
Resumo: O mundo dos negócios está mais competitivo do que nunca. Especificamente no caso de crédito bancário, possuir e utilizar ferramentas que possam auxiliar na tarefa de reconhecer e prever quais clientes serão "bons ou maus" tomadores de crédito, pode se tornar um fator chave, resultando numa grande vantagem competitiva. Existe muito conhecimento escondido na imensa quantidade de dados disponíveis nos bancos de dados das empresas. Com a metodologia de Data Mining, pode-se transformar esses dados em informações valiosas para auxiliar no processo decisório. Neste trabalho estão sendo analisados registros históricos de 339 clientes (pessoas jurídicas) de uma agência bancária, através de duas das ferramentas de Data Mining: Redes Neurais e Árvores de Decisão. Estas técnicas permitem fazer o reconhecimento de padrões e também diagnosticar novos casos. A idéia central deste trabalho é, portanto, utilizando as técnicas de Redes Neurais e Árvores de Decisão através do uso dos softwares MATLAB- Neurai Networks Toolbox e WEKA-Wa/Tcafo Environment for Knowledge Ânalysis, respectivamente, auxiliar na tomada de decisão sobre conceder ou não crédito bancário a um novo cliente. Os resultados foram bastante satisfatórios, mostrando que, para este problema específico, as Redes Neurais apresentaram um percentual menor de erros Abstract: The business world is more competitive than even Specificaíly in the case of bank credit, the use of toois that can assist in the task to recognize and to foresse which customers will be "good or bad" credit payers, can become a key factor resulting in a great competitive advantage. There is hidden knowledge in the immense available amount data in the databases of companies. With the Data Mining methodology, these data can be transformed into vaiuable Information to assist in the making decision process. Sn this work we analyzed historical registers of 339 customers of a bank agency, through two of the tools of Data Mining: Neuraf Networks and Decision Trees. These techniques alíow to make the pattern recognition and also to diagnose new cases. The central idea of this work is therefore, using the techniques of NeuraS Networks and DecisionTrees through the use of MATLAB-Neurai Networks Toolbox and WEKA-Waikato Environment for Knowledge Analysis softwares, respectively, to assist in the making decision on granting or not credit bank to a new customer. The results had been sufficiently satisfactory, showing that, for this specific problem, the Neural Nets had presented a lesser percentage of errors
Collections
- Dissertações [168]