A heuristic method for deriving range-based classification rules

The ability to learn classification rules from data is important and useful in a range of applications. While many methods to facilitate this task have been proposed, few can derive classification rules that involve ranges (numerical intervals). In this paper, we consider how range-based classification rules may be derived from numerical data and propose a new method inspired by classification association rule mining. This method searches for associated ranges in a similar way to how associated itemsets are searched in categorical attributes in association rule mining, but uses class values to guide the search, so that only those ranges that are relevant to the derivation of classification rules are found. Our preliminary experiments demonstrate the effectiveness of our method.