Article Details

Study on Association Rule Mining In Terms of Correlations Between Different Attributes |

Simerjit Kaur, Indu Singh, in Journal of Advances in Science and Technology | Science & Technology


Associationrule mining provides valuable information in terms of significant correlationsbetween different attributes’ values that might not be evident at the firstglance in large datasets. The experimental part of this work has demonstratedbenefits of integration of interactivity in Apriori approach for discoveringassociation rules hidden in the target dataset. The interactive algorithm fordiscovering association rules starts by asking user’s requirement with respectto attributes to be included in the search. Since the dataset has one classattribute that determines the patientclass (LIVE or DIE), the clinicians are interestedin finding rules that determine the value of patient class (LIVE or DIE). In addition to attributespecification, the user supplies the minimumsupport and confidencethreshold, the two parameters required by Apriori algorithm. In theexperimental runs, minimum supportand confidence threshold havebeen fixed at 15% and 80%, respectively