Hash Algorithm for Finding Associations between Genes

P. Asha¹ and S. Srinivasan²

¹Research Scholar, Computer Science and Engineering Department, Sathyabama University, Chennai, Tamilnadu, India. ²Professor and Head of the Department, Computer Science and Engineering, Anna University, Regional Centre, Madurai, Tamilnadu, India.

ABSTRACT: Association rules are those that narrate the relationships prevailing between attributes present in the database. Every rule mining algorithm generate promising items (frequent items) from which, the rules are framed. These rules try to state the items that are most related and how much one item is closer and depending on the other item. But the rules generated are enormous in number. Filtering out the useful patterns becomes difficult. The paper proposes a Hash based algorithm for extracting only the fruitful patterns at a faster rate. The work has been done using R language, and executed in R data mining Toolkit. Comparative study of Hash algorithm with respect to other algorithms shows that the Hash algorithm behaves better than all the other existing algorithms. It has been tested against various benchmark datasets like Adult, Genome, Cancer datasets using various rule interestingness measures like Lift, Confidence, Interest, Support etc.

KEYWORDS: Data mining; Associations; Rule Filtration; Interestingness Measures; Genes

[ HTML Full Text]