Nearby Neighboring Biclustering Approach

The ubiquitous use of mining techniques in today's era to mine germane pattern and information from the immense dataset has made the analysis of data an international priority. Aforesaid in literature, clustering is one of the most popular data mining techniques to knowledge discovery in data set. Clustering is the process of aggregating data into predefined k clusters based on some homogeneity criterion with the objective to maximize inter cluster distance and minimize intra cluster distance. Clustering is useful but is very inflexible in some scenario which yields an approach biclustering. A study of biclustering approach is useful to unfold the hidden coherent patterns from the data matrix. Biclustering accomplishes simultaneous pattern classification in both row and column directions in the data matrix, which differs to traditional clustering procedure that searched for aggregation of coherent patterns using the entire feature set. It is a process to extract subsets of objects over a subset of attributes from high dimensional datasets. In this paper, we present a novel approach, called NBM to discover the biclusters. Our proposed schemes encompass two phases. The first phase of the novel approach generate clusters based on a nearby neighbor. In the second phase, the biclusters are created by accomplishing the fusion operation on the clusters group in some manner. The paper introduces a novel approach that we feel can contribute significantly in the endeavor to create better and more efficacious biclusters.