Malaysian Journal of Mathematical Sciences, September 2018, Vol. 12, No. 3
Application of Classification and Regression Trees Algorithm to Classify Children Ever Born: BDHS 2011
Saadati, M., Bagheri, A., and Rana, S.
Corresponding Email: abagheri_000@yahoo.com
Received date: 4 July 2017
Accepted date: 30 August 2018
Abstract:
Study the nature of fertility determinants as influential development indicators is a necessity in densely populated countries like as Bangladesh. In addition, investigating these factors without considering the convenient statistical methods may result in misleading conclusions. The main purpose of this article is to classify one of the most important principal of fertility, children ever born, by applying Classification and Regression Trees (CART) algorithm. To achieve this goal, children ever born of ever-married women age 12-49 years old from Bangladesh Demographic and Health Survey 2011 data has been classified by Classification and Regression Trees algorithm according to a number of candidate demographic and socio-economic predictors. Marriage duration, couple’s educational level, division, and religion were determined to be the most influential predictors by extracted classification model. The efficiency of CART algorithm has been proved by accuracy of the model.
Keywords: Fertility, Children Ever Born, Decision Trees, Regression Trees, Bangladesh Demographic and Health Survey
