Rule Power Factor: A New Interest Measure in Associative Classification
Ochin, Suresh Kumar V, Nisheeth Joshi · Procedia Computer Science · 2016
In data mining it is generally anticipated that revealed knowledge should have characteristics of accuracy, reliability and interestingness. Most of the data mining algorithms find patterns that are accurate and reliable but might not be interesting. Interest measures are used to find the valued interesting rules which are useful to the user in effective decision making even in exceptional set of circumstances. A range of interest measures for rule mining have been suggested by researchers in the field of data mining to have different visualisations and analytics. In this paper, we have investigated a few of interest measures and proposed a new Interest measure with the name ‘Rule Power Factor’. Experiments prove that this new interest measure is more informative and can act as a superset of ‘Confidence’ measure.