Resumen

PEREZ-SALVADOR, Blanca Rosa; SANCHEZ-LORDMENDEZ, Carlos Gabriel  y  BAPTISTA-GONZALEZ, Héctor Alfredo. Binary classification to predict familiar thrombosis. Ing. invest. y tecnol. [online]. 2017, vol.18, n.4, pp.457-464. ISSN 2594-0732.

Thrombosis and other health problems can be difficult to diagnose. The diagnosis is a classification problem. In this work a classification problem is solved using free software on a database of patients from the National Institute of Perinatology Isidro Reyes Espinoza (Instituto Nacional de Perinatología Isidro Espinoza Reyes) with thrombosis problems. In this work the risk that a person will develop the disease mentioned was analyzed. The study includes, variables selection, imputation and analysis of two classification methods: one based on logistic regression and the other in the likelihood ratio. The results indicate that the likelihood ratio classification was better than logistic regression classification.

Palabras llave : binary classification; logistic regression; Neyman-Pearson lemma; deviance.