Parallel Mining of Frequent Item sets using Map Reduce Technique: A Survey
Prajakta G. Kulkarni1, Rubeena A. Khan2
1Prajakta G. Kulkarni, Department of Computer Engineering, Modern Education Society’s College of Engineering, Savitribai Phule Pune University, Pune (Maharashtra) – 411001, India.
2Prof. Rubeena A. Khan, Department of Computer Engineering, Modern Education Society’s College of Engineering, Savitribai Phule Pune University, Pune (Maharashtra) – 411001, India.
Manuscript received on November 25, 2018. | Revised Manuscript received on November 28, 2018. | Manuscript published on January 30, 2018. | PP: 1-5 | Volume-6 Issue-6, January 2017. | Retrieval Number: F2934016617
Open Access | Ethics and Policies | Cite | Mendeley
© The Authors. Published By: Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: In the task of data mining, the most important job is to find out frequent itemsets. Frequent itemsets are useful in various applications like Association rules and correlations. These systems are using some algorithms to find out frequent itemsets. But these parallel mining algorithms lack some features like automatic parallelization, well balancing the load, distribution of data on large number of clusters. So there is a need to study the parallel algorithms which will overcome the disadvantages of the existing system. In this paper a technique called fidoop is implemented, In this technique the mappers work independently as well as concurrently. This is done by decomposing the data across the mappers. Reducers work is to combine these jobs by developing small ultra metric trees. To show this fidoop technique on the various clusters is very delicate in distribution of data because different datasets are with different partition of data. This fidoop technique is also useful in heterogeneous clusters.
Keywords: Frequent item sets, mappers, reducers, Ultrametric trees, FiDoop.