Research Of Multi-Dimensional Numeric Association Rules

In recent years, data mining catches a great deal of attention in the information industry. The main reason of it is due to the wide availability of huge amounts of data and the imminent need for turning such data into useful information and knowledge.Mining association rules is one of the best important areas of data mining, which is a process of extracting interesting and frequent patterns from the data set. There are many classic algorithms on mining association rules, such as Apriori. These algorithms mostly imply in data, which is a type of Boolean. Because data type in database is richer, mining on data type except for Boolean is one of major topic in association rules mining.Generally, data type in database can be quantitative or categorical. The former is more important than latter. For numeric data, the key step is dividing the value domain into intervals. In this paper, we combine cluster method with mining association rules. We propose a concept INTERVALS which describe the scope of interval. Based on the concept, we divided the value domain into intervals.This paper is organized as follow. First we introduce data mining. Then we systematic research on classic association rules mining algorithms. Theses algorithms are base of multi-dimension association rules mining. There are two kind algorithms on multi-dimension association rules mining, merging intervals and clustering. We introduce those respectively. In this paper, we propose an algonthmCARB, which is a one-pass algorithm. While user specify INTERVALS , the algorithm CARB can divided values into intervals.Our research can contribute on multi-dimension association rules and data pre-process.