Loading...
Modelo para el análisis de los factores asociados con el tipo de parto aplicando bosques aleatorios y regresión logística
López Limas, Lesbia O.
López Limas, Lesbia O.
Citations
Altmetric:
Abstract
Actualmente existe una variedad de algoritmos que tratan de optimizar el proceso de clasificación, entre ellos el algoritmo de Bosques Aleatorios. Uno de los objetivos de esta investigación es desarrollar un modelo de clasificación para el tipo de parto usando Bosques Aleatorios. El modelo de clasificación mediante Bosques Aleatorios ha sido aplicado a datos de nacimientos en Puerto Rico para el año 2,017 con el objetivo de clasificar el tipo de parto (vaginal o cesárea) de una mujer en estado de embarazo de acuerdo a características socio-demográfcas, previas al embarazo, durante el embarazo y relacionadas con el neonato. El método de Bosques Aleatorios se utilizó además, para determinar las puntuaciones de importancia de las variables predictoras en la clasifcación del tipo de parto, con el fin de realizar una selección de variables para predicción. Con estas variables de predicción, se elaboró el modelo de regresión logística binaria, para cuantifiar el efecto de dichas variables en el tipo de parto. En particular se verificó que estas variables eran factores de riesgo para el parto por cesárea y se cuantificó cada uno de los Odds ratio asociados a las mismas.
There are currently a variety of algorithms that try to optimize the classification process, including the Random Forest algorithm. One of the objectives of this research is to develop a classification model for the type of delivery using Random Forest. The classification model through Random Forest has been applied to birth data in Puerto Rico for the year 2,017 with the objective of classifying the type of delivery (vaginal or caesarean section) of a woman in pregnancy according to sociodemographic characteristics, before pregnancy, during pregnancy and related to the newborn. The Random Forest method was also used to determine the importance scores of the predictive variables in the classification of the type of delivery, in order to make a selection of variables for prediction. Using these prediction variables, a binary logistic regression model was developed to quantify the e ect of these variables on the type of delivery. In particular, it was verified that these variables were risk factors for cesarean delivery and each of the Odds ratio associated with them was quantified.
There are currently a variety of algorithms that try to optimize the classification process, including the Random Forest algorithm. One of the objectives of this research is to develop a classification model for the type of delivery using Random Forest. The classification model through Random Forest has been applied to birth data in Puerto Rico for the year 2,017 with the objective of classifying the type of delivery (vaginal or caesarean section) of a woman in pregnancy according to sociodemographic characteristics, before pregnancy, during pregnancy and related to the newborn. The Random Forest method was also used to determine the importance scores of the predictive variables in the classification of the type of delivery, in order to make a selection of variables for prediction. Using these prediction variables, a binary logistic regression model was developed to quantify the e ect of these variables on the type of delivery. In particular, it was verified that these variables were risk factors for cesarean delivery and each of the Odds ratio associated with them was quantified.
Description
Date
2019-12-11
Journal Title
Journal ISSN
Volume Title
Publisher
Collections
Keywords
Random Forest, Logistic regression analysis