Why Imprecise regression: a discussion
Résumé
Machine learning, and more specifically regression, usually focuses on the search for a precise model, when precise data are available. It is well-known that the model thus found may not exactly describe the target concept, due to the existence of learning bias. In order to overcome the problem of learning models having an illusory precision, a so-called imprecise regression method has been recently proposed for non-fuzzy data. The goal of imprecise regression is to find a model that offers a good trade-off between faithfulness w.r.t. data and (meaningful) precision. In this paper, we propose an improved version of the initial approach. The interest of such an approach with respect to classical regression is discussed in the perspective of coping with learning bias. This approach is also contrasted with other fuzzy regression approaches.