Pages: 35-41

Abstract

Abstract: Decision tree as one of data mining technique can validate bank customers. Main issue is the construction of a decision tree that can optimally classify customers. In thesis, presented a suitable model validation bank customers for providing credit facilities proportionate with each class based on genetic algorithm. Genetic algorithms by choosing the right features and making the optimal decision tree can help validation of customers. In pattern recognition and used process for CRISP validation of customers. The proposed classification model is based clustering techniques, characters selection, decision trees and genetic algorithm. This model tries to select and combine the best decision trees based on optimal standards and making the final decision tree to validate customers. The results show that accuracy of classification of model proposed classification is higher than all the decision tree models in this thesis. Also number of leaves and the size of decision tree and also the complexity of it are lower than all the other aspects of it.