Analysis of Medical Domain Using CMARM: Confabulation Mapreduce Association Rule Mining Algorithm for Frequent and Rare Itemsets

Abstract

Abstract-In Human Life span, disease is a major cause of illness and death in the modern society. There are various factors that are responsible for diseases like work environment, living and working conditions, agriculture and food production, housing, unemployment, individual life style etc. The early diagnosis of any disease that frequently and rarely occurs with the growing age can be helpful in curing the disease completely or to some extent. The long-term prognosis of patient records might be useful to find out the causes that are responsible for particular diseases. Therefore, human being can take early preventive measures to minimize the risk of diseases that may supervene with the growing age and hence increase the life expectancy chances. In this paper, a new CMARM: Confabulation-MapReduce based association rule mining algorithm is proposed for the analysis of medical data repository for both rare and frequent itemsets using an iterative MapReduce based framework inspired by cogency. Cogency is the probability of the assumed facts being true if the conclusion is true, means it is based on pairwise item conditional probability, so the proposed algorithm mine association rules by only one pass through the file. The proposed algorithm is also valuable for dealing with infrequent items due to its cogency inspired approach.

Prognosis of Disease that may Occur with Growing Age using Confabulation Based Algorithm

“…The proposed algorithm is designed as an iterative map reduce procedure based on confabulation method 20,22,23 . It is processed in two steps: 1.…”

“…This leads to the generation of frequent and rare itemset, whose support count is greater than the minimum specified support count value. In second step, strong association rules are generated based on confabulation theory, which states 20 ;…”

“…It considers S 1 = F r where S 1 is set of initial itemset. After finding n-itemset, algorithm generates all rules using their support, confidence and interestingness parameter for selecting active items and interesting rules that strengthen the knowledge link between the itemsets 13,20 .…”

“…The CMARM approach is divided into three modules 20,22,23 ; (i) Creating distributed cluster: Since, the medical data deals with voluminous and heterogeneous data structure. Therefore, to handle such big amount of data HDFS is used 3,10,14 .…”