An improved K means clustering with Atkinson index to classify liver patient dataset

In data mining or machine learning clustering is very broad area. Clustering is a technique which decomposes the data set into different cluster. There are many clustering algorithms but k-mean algorithm is most popular and widely used in many fields such as image processing, machine learning, pattern reorganization etc.; but it has a major drawback that is; its output is really sensitive to the random selection of initial centroids or its final output is totally depends on initial selection of centroids. Because of this drawback many techniques were introduced for K-mean algorithm. This paper introduces an initial centroid selection method for K-mean algorithm by using Atkinson Index. Atkinson index is a technique for measuring the inequality here. It is used for initial seed selection; experimental result shows that the proposed technique, which we applying on liver patient data set gives more accurate result than the original K-mean algorithm.