| Original language | English |
|---|---|
| Title of host publication | Encyclopedia of machine learning |
| Editors | Claude Sammut, Geoffrey I. Webb |
| Publisher | Springer New York |
| Pages | 938–941 |
| ISBN (Electronic) | 978-0-387-30164-8 |
| ISBN (Print) | 978-0-387-30768-8 |
| DOIs | |
| State | Published - 2010 |
Abstract (may include machine translation)
Supervised descriptive rule induction (SDRI) is a machine learning task in which individual patterns in the form of rules (see Classification rule) intended for interpretation are induced from data, labeled by a predefined property of interest. In contrast to standard supervised rule induction, which aims at learning a set of rules defining a classification/prediction model, the goal of SDRI is to induce individual descriptive patterns. In this respect SDRI is similar to association rule discovery, but the consequents of the rules are restricted to a single variable – the property of interest – and, except for the discrete target attribute, the data is not necessarily assumed to be discrete.