FARM: a new efficient and effective data clustering algorithm
Cheng-Fa Tsai, Kuei-Sheng Lee · 2009
This investigation presents a method named FARM that combines a grid-based algorithm with the density-based approach for clustering data in data mining applications. In the FARM clustering method, the number of separate clusters need not be specified but only the number of divisions of the clusters is required. Experimental results indicate that the proposed method clusters correctly. It filters 98.8% of the noise, and the data set accuracy exceeds 99.7%. The most surprising result is the time required to process data sets. Processing 575,000 data sets takes only 0.33 second - much less time than any currently known clustering algorithm.