In the recent years, there is an increasing interest of the use of case-based reasoning (CBR) in medicine. CBR is an approach
to problem solving that is able to use specific knowledge of previous experiences. However, the efficiency of CBR strongly
depends on the similarity metrics used to recover past experiences. In such metrics, the role of attribute weights is critical.
In this paper we propose a methodology that use subgroup discovery methods to learn the relevance of the attributes. The methodology
is applied to a Breast Cancer dataset obtaining significant improvements. ...