S.

No

Methodology in

Proposed System

Short Description

Advantage

1.

Agglomeration

・ Every data objectis is assigned to cluster only once.

・ No reallocation.

Fast merging with

time complexity O(N²)

2.

Exact Similarity Measure &

No Approximation

・ Outliers are eliminated automatically during agglomeration.

・ Elimination of noise

Improvement

in cluster validity

index value

3.

Preprocessing

・ Uninteresting transactions are

・ eliminated.

・ Deviation in knowledge extraction is constrained.

Interesting patterns are identified

4.

Threshold

・ Setting proper threshold removes

・ distant objects

Improvement in cluster validity

5.

Merging Closest Pair

・ Merge step ends automatically. Therefore no need to predefine k

No need to specify number of clusters