Quantitative association rules problem is to discover associations among quantitative attributes. Usually, the approach is to map this problem into the Boolean association rules problem. The quantitative value is first partitioned into intervals and then each pair is mapped into a Boolean attribute. Whether to partition a quantitative attribute and how many partitions there should be are the main problem of the mapping. In this paper, we present a clustering algorithm for partitioning quantitative values and a combining algorithm for obtaining a simple rule set. Our method utilizes the distribution of the transactions for each attribute value. It does not need a predefined threshold and only considers those interesting intervals.
展开▼
