Designing a New Fuzzy Genetic Gravity Algorithm for Data Mining

Nowadays, due to the high volume of data and their complexity and the human needs to the hidden knowledge in them, using an efficient method is necessary. In this study, our purpose is to present an efficient method of data mining in order to fetch knowledge of input data set. The given proposed method - with the help of fuzzy systems based on rules which are a set of if-then rules - fetch the needed knowledge and classify it. Fuzzy rules are paid attention because they can be interpreted by a human expert. In fact, the our purposed knowledge can be considered as a fuzzy database which improved during the data mining process and with the help of optimum algorithm according to the criteria such as accuracy and ability to interpretation. In order to optimize the obtained fuzzy rules set, a combination of genetic algorithm and the assimilated cooling heuristic is used. The assimilated cooling which is based on statistical mechanics, according to the criteria of accuracy and ability to interpret, try to find a set of fuzzy if-then rules in the state space related to set of rules which can have the best performance and that can escape from the local optimal solutions with the help of genetic algorithm mechanisms. Finally, the proposed method has been implemented in software and applied on a set of UCI data set. The obtained results were compared with the results of the famous methods in this field such as Support vector, N-bayes, KNN , D-tree, GBML and they have shown good accuracy and efficiency.