Data mining application to optimize time and memory utilization using distributed clustering methods for Large multidimensional databases with dynamic index and search structures

Abstract

Clustering algorithms has emerged as an important data mining technique for pattern recognition, data analysis and dimensionality reduction. In recent times clustering is being used as a basic data analysis technique in different analytic applications, but understanding and evaluating multidimensional clustering results have become a complicated task for many users particularly regarding the quality of the clusters and their semantics. Highest level of statistical information is needed about the clusters when we deal with large and complicated data. The statistical information facilitates the newlineusers to evaluate the cluster quality along with a detailed display of multidimensional attributes which help the users to understand the meaning of the clusters. There are many techniques and algorithms for effective clustering process based on search structure but those algorithms have only limited success due to complexity of large data and multidimensional indexes. To solve this issue, effective clustering methods for large data are considered and implemented. These methods are good in their results but still underperform in terms of time effectiveness and accuracy of the data. In order to overcome the above limitations, effective clustering methods such as Distributed newlineWeighted Possibilistic C-Means algorithm (DWPCM) and Distributed Weighted Fuzzy C-Means algorithm (DWFCM) are proposed in this research. newline newline