An adaptive rough fuzzy single pass algorithm for clustering large data sets

Cluster analysis has been widely applied in many areas such as data mining, geographical data processing, medicine, classi-cation of statistical -ndings in social studies and so on. Most of these domains deal with massive collections of data. Hence the methods to handle them must be e/cient both in terms of the number of data set scans and memory usage. Several algorithms have been proposed in the literature for clustering large data sets viz; CLARANS [1], DB-SCAN [1], CURE [1], K-Means [2], etc. Most of these require more than one pass through the data set to -nd the required abstraction. Hence they are computationally expensive for the clustering of large data sets. Even though we have a single pass clustering algorithm called BIRCH [1], it uses a memory expensive data structure called CF tree. In this scenario the Leader algorithm [3], which requires only a single data set scan and less memory, turns out to be a potential candidate. This paper introduces an e/cient variant of leader algorithm called Adaptive Rough Fuzzy Leader (ARFL) algorithm which out-performs the conventional leader algorithm. It employs a combination of Rough Set theory [4] and Fuzzy set theory [5] to capture the intrinsic uncertainty involved in cluster analysis. The paper is organized as follows. Section 2 discuss the conventional Leader algorithm, Section 3 introduces the proposed algorithm, in Section 4 a comparative study of the algorithm with conventional leader algorithm and single pass K-means algorithm is given and Section 5 deals with conclusions.