Obtaining a Linguistically Understandable Random Sets-Based Classifier from Interval-Valued Data with Genetic Algorithms

Abstract

Combining descent algorithms and a coevolutionary scheme, we have defined a new procedure that is able to obtain rule-based models from datasets with censored or interval-valued data, and can also identify the conflictive instances in the training set: those that contribute the most to the indetermination in the likelihood of the model.