Intelligent hybrid system for data mining

Summary form only given. Database mining is the process of finding patterns and relations in large database. A number of database mining techniques have been developed in domains that range from space and ocean exploration to financial and business analysis. The models generated from using data mining processes are statistical (e.g., linear regression, and nonlinear regression), symbolic (e.g., decision tree, CART, ID3), fuzzy symbolic (fuzzy logic systems), neural (feedforward neural network, recurrent neural networks, and self-organizing memory SOM), and genetic (genetic algorithm based on the biological survival of the fittest). Some scientists are trying to introduce chaos theory and fractal statistics for better data mining. It is the conflict between the symmetry of the Euclidean geometry and the asymmetry of the real randomness and determinism, chaos and order coexist. While these intelligent techniques have produced encouraging results in particular tasks, certain complex problems cannot be solved by a single intelligent technique alone. Each intelligent technique has particular computational properties that make them suited for particular problems. These limitations have been a central driving force behind the creation of intelligent hybrid systems. For example, the combination of neural network and fuzzy logic systems has been applied successfully in loan evaluation, fraud detection, financial risk assessment, financial decision making, and credit card application evaluation. We present a novel hybrid system for data mining in financial analysis.