Labeled clustering a unique method to label unsupervised classes

摘要

This paper proposes a method to label unsupervised classes. Clustering is an unsupervised classification technique that is used to group data on the basis of similarity measures. K-Means clustering is one of the methods which is used to classify given dataset in number of groups on the basis of Euclidean distance of data points in Cartesian system. On the basis of similarity and dissimilarity the data points are divided into multiple clusters which do not have identification labels. K means clustering can given a broadened insight into the data if the resulting groups bear some identification. A unique method for labeling unsupervised classes by using correlation analysis and frequent membership function is proposed. The method is applied to custom world energy dataset and divided world nations into five labeled clusters which increased the opportunities for energy sector to derive valuable patterns for guided decisions. The results showed minor deviations from the real energy scenario because of the factors discussed in the paper.