A High Dimensional Clustering Scheme for Data Classification

The data mining is the knowledge extraction or finding the hidden patterns from large data these data may be in\ud different form as well from different resources. The data mining systems can be used in various research\ud domains like health, share market analysis, super market, weather forecasting and many other domains. Data\ud mining systems use the computer oriented algorithms. These algorithms can be categorized as supervised and\ud unsupervised respectively. The classification or prediction algorithms belong to supervised category and\ud clustering algorithms are the type of unsupervised. The clustering is an approach to group the similar data\ud objects to the same cluster and the main aspect of clustering is that the distance between data objects in same\ud cluster should be as minimum as and distance of objects in inter cluster should be high. k-means is one of the\ud most common clustering algorithm. K-means is very easy to use and efficient but has also some weakness\ud because of random or inappropriate selection of initial centroids so need to improve k-means. The proposed\ud work is an attempt to improve k means by using genetic algorithm for selection of initial cluster centroid