Open AccessBivariate zero inflated generalized Poisson regression model in the number of pregnant maternal mortality and the number of postpartum maternal mortality in the Central Java Province in 2017
Author(s) -
Qurotul Aini,
Purhadi Purhadi,
Irhamah
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1511/1/012055
Subject(s) - poisson regression , statistics , mathematics , population , poisson distribution , linear regression , bivariate analysis , regression analysis , medicine , demography , environmental health , sociology
Excess zero is one of the problems in Generalized Poisson regression where the number of responses is contained zero exceeds 60 percent. One of the statistical methods have been developed is Zero Inflated Generalized Poisson Regression (ZIGPR). If there are two response variables, the appropriate regression analysis is Bivariate ZIGPR (BZIGPR). This study aims to determine the factors that influence the number of pregnant maternal mortality and the number of postpartum maternal mortality in 91 sub districts in Pekalongan Residency, Central Java Province 2017 through by the BZIGPR. The variables are percentage of K1 pregnancy examinations, percentage of K4 pregnancy examinations, percentage of deliveries assisted by health workers, percentage of pregnant women who received Fe3, percentage of TT2 + immunization in pregnant women, ratio of midwives per 100,000 population and percentage of obstetric complications management. The estimation of BZIGPR parameters using the Maximum Likelihood Estimation (MLE) method which results in a non-linear form so that, it is solved by the Berndt Hall-Hall Hausman (BHHH) iteration method. The result hypothesis testing using the Maximum Likelihood Ratio Test (MLRT) is reject null hypothesis. BZIGPR produces 2 regression models ln μ ^ l i and logit p ^ l i . The predictor variables that affect response Y 1 and Y 2 on model ln μ ^ l i are all of predictor variables. The predictor variables that affect response Y 1 on model logit p ^ l i are all of predictor variables while the predictor variables that affect response Y 2 are all of predictor variables except percentage of pregnant TT2 + immunization women and percentage of midwives per 100,000 population.