Mensagem "A common criticism to decision trees is that once the training set is divided after answering a question, it is not pos…

"A common criticism to decision trees is that once the training set is divided after answering a question, it is not possible to reconsider this decision. For example, if we divide men and women, every subsequent question would be only about men or women, and the method could not consider another type of question (say, age less than a year, irrespective of the gender). Random Forests try to introduce some level of randomization in each step, proposing alternative trees and combining them to get the final prediction. These types of algorithms that consider several classifiers answering the same question are called ensemble methods." Do livro Learning Scikit Learn, por Raúl Garreta and Guillermo Moncecchi , 2013. Publicado Na Packtpub