Discovery algorithms for hierarchical relations

摘要: Two algorithms based on a latent class model are presented for discovering hierarchical relations that exist among set ofK dichotomous items. The two algorithms, stepwise forward selection and backward elimination, incorporate statistical criteria selecting (or deleting) 0–1 response pattern vectors to form the subset of total possible 2 k uniquely describe hierarchy. performances compared, using computer-constructed data, with those three competing deterministic approaches ordering theory calculation Phi/Phi-max coefficients. discovery also demonstrated real data sets investigated in literature.