Typical Testor Selection Process for Classification Models

This work develops a process based on Testor Theory and new ideas related to this field that allows a ranking of typical testors based on the similarity between objects from the same class and the dissimilarity between objects from different classes. This process allows us to select the typical testors that will perform effectively to reduce the number of features in a dataset. We validate this process by examining the results obtained from a classification model for three different datasets, using the best and worst-ranked typical testors selected by the algorithm. Lastly, we analyze the results obtained and present the effectiveness of the method, as well as the advantages it brings.