A Guided Clustering Technique For Knowledge Discovery

Improving Data Mining Performance through Scalable Guided Clustering Technique

Abstract

Scalability: As visualization techniques are used more andmore in data mining tools, it becomes important to address scalability issue.Fundamentally, the reason for scalable solutions is to be able to build a gooddata mining model as quickly as possible. This offers two benefits, a) there isvalue in being able to deploy and use the model sooner rather than later and b)faster turnaround times yield better models. Time previously spent waiting forresults can instead be devoted to finding the model that result in the best andmost reliable solution. Making data mining tools scalable requires hardwarescalability and parallel algorithms. The goal of hardware scalability is toprovide high performance by adding modestly priced processor building blocks(or nodes) in such a way that performance scales linearly.