Combining Techniques to Find the Number of Bins for Discretization
Miguel Á. Álvarez‐Carmona, Jesús Ariel Carrasco-Ochoa, José Fco. Martínez-Trinidad · 2013
In different problems and for different reasons, it is necessary to work with fully discretized domains. Commonly, continuous attributes are discretized in bins, but determining a suitable number of bins is a difficult task. In this paper, some ways for combining different techniques to estimate the number of bins for discretization are proposed and evaluated for both clustering and supervised classification tasks.