Mining Multilevel Association Rule at Different Concept Hierarchy

Data warehouses receive huge amounts of data from a variety of sources which may contain “noisy data” and is used in decision making. Data mining is extracting knowledge from huge amount of data usually contain some amount of missing data along with a variable percentage of inaccurate data, pollution, outliers and noise. The actual data mining process deals significantly with prediction, estimation, classification, pattern recognition and the development of association rules.One way of detecting data errors is by utilizing association rules which specify relationships between record attributes. Association rule algorithms identify patterns that occur in the large database. Apriori like candidate-generation-and-test approach may encounter serious challenges when mining datasets with long patterns. Mining association rules at multiple concept levels may lead to the discovery of more specific and concrete knowledge from data. The discovery of multiple level association rules is very much useful in many applications. The paper mentions the need for computer applications that implement algorithms based on multiple-level association rule mining. WEKA, developed at the University of Waikato in New Zealand, is a computer application that is widely used for data mining. In this paper, describes the implementation of multilevel association rule mining efficiently using concept hierarchies.