Design and analysis of effective partitioned based clustering algorithm and its application

Abstract

These days a numerous data are generated and gathered by a mixed bag of disciplines In that respect there is a need to use these available data with efficiency and convert them into valuable information for the decision making process To achieve this goal one needs data accumulation analytic thinking and evaluation process Generally this process is called knowledge discovery and as the data is stored in a database they are also recognized as knowledge discovery in databases or data mining newlineThe machine learning community over the last decade continually striving to create computational methods for exploiting the knowledge The machine learning community uses two types of learning process that covers the major algorithm present in this field Supervised learning where information about the data is available and unsupervised learning where information about the data is not available In unsupervised learning only the similarity between the data objects can be exploited in order to discover groups of data that share some basic structure newlineIn order to describe and characterize objects by their properties in dataset and in exploratory data analysis grouping is an important technique Clustering algorithm is applied in a variety of scientific as well as engineering applications to organizes data on the basis structure of data objects either by grouping individual data objects or as a hierarchy of the groups It is also called as unsupervised learning in the field of artificial intelligence and pattern recognition newlineThe k means algorithm is among the topmost algorithms in the field of data mining and machine learning due to its simplicity and ease in implementation Despite being discovered in 1967 it remains popular clustering algorithm till date This algorithm begins with picking k initial seed values which can be used as initial centres The selection of initial set of centers is critical as it has a great impact on the quality and speed of the algorithm Therefore finding a way for initialization the k-means algorithm