A novel fuzzy association rule for efficient data mining of ubiquitous real-time data

Abstract

In ubiquitous stream of data, the issue related to the association rules of fuzzy are considered in this paper, and a new method FFP_USTREAM (Fuzzy Frequent Pattern Ubiquitous Streams) are created. The system of Ubiquitous real-time data incorporates fuzzy ideas with automated streams of data, utilizing the method of sliding window, to mine rules associated for fuzzy logic. The proposed strategy used a matrix of fuzzification where the input patterns related to level of membership to various classes. Attribution of specific classification or class is depending on estimation level of pattern membership. This technique is applied to ten benchmarks data set with classification of learning repository from the UCI machine. The motivation is to evaluate the proposed strategy and, in this manner the performance is compared to a pair of incredible supervised classification algorithms sigmoidal Recurrent Neural Network (RNN) and Adaptive Neuro-fuzzy Inference System (ANFIS). An efficient and complexity of the system are examined. Instances of genuine set of data are utilized to test the proposed system. Existing regression and classification methods is used to compare the proposed fuzzy method. Proposed fuzzy achieves better results when compared to existing method.