(Insert Figure 3 about here)
The generation mechanism divides the dataset into several subsets. In the proposed
algorithm, this mechanism employs different algorithm to avoid similar subsets. There are
four basic clustering algorithms applied, namely, K-means, single linkage agglomerative
hierarchical clustering (SLHC), complete linkage agglomerative hierarchical clustering
(CLHC) and average linkage agglomerative hierarchical clustering (ALHC). From these
four algorithms, three different combinations are proposed. They are: (1) K-means
algorithm, (2) SLHC, CLHC and ALHC, and (3) K-means, SLHC, CLHC and ALHC. The
subsets are also generated from different attributes, , and instances
. The combinations of attributes and instances are generated randomly. Figure
4 illustrates the subsets generation.