MR_IMQRA: An Efficient MapReduce Based Approach For Fuzzy Decision Reduct Computation

EasyChair Preprint 5074

Abstract

Fuzzy-rough set theory, an extension to classical rough set theory, is effectively used for attribute reduction in hybrid decision systems. However, it's applicability restricted to moderate size data sets because of higher space and time complexities. In this work, an algorithm MR\_IMQRA is developed as a MapReduce based distributed/parallel approach for standalone fuzzy-rough attribute reduction algorithm IMQRA. The proposed approach is developed for scalability in attribute space and is relevant for scalable attribute reduction in the areas of Bioinformatics and document classification. This algorithm uses a vertical partitioning technique to distribute the input data in the cluster environment of the MapReduce framework, which reduces the complexity of data movement in shuffle and sort phase of MapReduce framework. The absolute positive region removal aspect of IMQRA is successfully incorporated in MR\_IMQRA so that the algorithm's computational efficiency is further improved. A comparative experimental analysis is conducted on larger attribute space hybrid decision systems, and the results demonstrated that the proposed MR\_IMQRA algorithm had obtained reduct in less computational time with good sizeup and speedup performance.

Keyphrases: Apache Spark, Fuzzy rough sets, Hybrid decision systems, Iterative MapReduce, Vertical partitioning, attribute reduction

@booklet{EasyChair:5074,
  author    = {Kiran Bandagar and Pandu Sowkuntla and Salman Abdul Moiz and P. S. V. S Sai Prasad},
  title     = {MR_IMQRA:  An Efficient MapReduce Based Approach For Fuzzy Decision Reduct Computation},
  howpublished = {EasyChair Preprint 5074},
  year      = {EasyChair, 2021}}