Comparison of clustering algorithms using quality metrics with invariant features extracted from plant leaves

This paper presents evaluation of the performance of clustering algorithms like Fuzzy C Means, Agglomerative and CURE in conjunction with cluster quality metrics namely Purity, Inverse Purity, Homogeneity, Completeness, Rand Index, V measure, Precision, Recall, F measure, Jaccard Coefficient and Folkes and Mallows. The effectiveness of the different quality metrics and clustering methods evolving the appropriate number of clusters is demonstrated experimentally for leaf data set with the number of clusters varying from five to fifteen. Once the appropriate number of clusters is determined, the performances of all clustering techniques are evaluated for appropriate grouping of the data into the number of clusters.