Learning weights for weighted OWA operators

Weighted OWA (ordered weighted aggregation) operators were introduced as a generalization of the weighted mean (WM) and the OWA operators, so that the advantages of both could be used in a single data fusion function. In this work, we study the determination of their parameters when a set of examples is at our disposal. The approach presented in this paper is of interest in data mining when a certain variable has to be expressed in terms of some other ones. In this case, the learning of weights corresponds to fitting the model, and the weights correspond to the importance of the variables and of their values.