Data Mining Applications And Optimization Studies

Data mining is a new technology that developing in recent years. By the use of data mining, people can find potentially valuable knowledge, which strongly supports the decision making in business scientifically. While the technology is rapidly developing, as an important part of data mining, clustering analysis is always being paid great attention. On the other hand, data visualization is a highly active topic in the data mining research as well.Firstly, this thesis introduces the related background of data mining and its theories. Then it briefly summarize the related work of clustering analysis. The thesis made a profound research on clustering, and processing in following innovative achievements:(1) Proposing a fuzzy clustering algorithm with adaptive parameterωThis thesis optimized the convenience of the fuzzy clustering algorithm on fuzzy equivalence relations which is a basis of data mining application,proposes a fuzzy clustering algorithm with adaptive parameterω. It avoids the complexity of manually setting of various parameters and makes the calculation more automatic. On the other hand, it calculates samples that with different properties directly without pre-conversion and thus expands the scope of application.(2) Proposing a technique that can trace the reverse processing in data visualization based on multidimensional grid clusteringData mining visualization refers to a technique that displays all parts of processing, but the multi-tiered computing and a series of data conversion and clean-up, the reverse visual analysis becomes a great problem. In this thesis, we use the particularity of grid computing to establish a forward and reverse mapping table with mapping formulas to achieve a small grid data independently, and result in realization of reverse visual analysis on a single point and makes the analysis more relevant.