Yazar "Alnakawa, Mohamad" seçeneğine göre listele

Listeleniyor 1 - 1 / 1
  • Küçük Resim
    Öğe
    An application based on zero-inflated poisson and negative binomial regression models
    (Selçuk Üniversitesi Fen Bilimleri Enstitüsü, 2020) Alnakawa, Mohamad; İyit, Neslihan
    In generalized linear models (GLM), the distribution of the response variable is considered to come from the exponential family. In GLM, besides the exponential family, the distribution of the response variable can also come from a zero-inflated Poisson and Negative-Binomial distributions. In this thesis study, after general information about GLM structure is given, Poisson and Negative Binomial Regression Models, which are the special cases of GLM, are introduced as in terms of model building processes and parameter estimates. Then, Zero-Inflated Poisson (ZIP) and Zero-Inflated Negative Binomial (ZINB) Regression Models are given in detail. The goal of this study is to determine the possible risk factors affecting "child mortality under-5" in some lower middle-income countries such as Cameron, Congo, Sudan, Vietnam, Pakistan (Punjab region), Pakistan (Sindh region), Zimbabwe, Kyrgyzstan, El Salvador and Mongolia by using the "Multiple Indicators Cluster Survey 5 (MICS5)" conducted by UNICEF in 2014, by using Poisson regression model, negative binomial (NB2) regression model, zero-inflated Poisson (ZIP) regression model and zero-inflated negative binomial (ZINB) regression model in the class of generalized linear models (GLMs). For this purpose, education (as attended to a school or not attended to a school), wealth index quintile (as poorest, second, middle, fourth and richest population wealth quintiles), area of residence (as rural or urban), mother's age (as 15-19, 20-24, 25-29, 30-34, 35-39, 40-44, 45-49 years), breastfeeding status (as received breast milk or not received breast milk) and pregnancy method (as used any methods to avoid pregnancy or not) are taken as categorical independent variables as possible risk factors affecting "child mortality under-5". Thus, considering the socio-economic and demographic characteristics of the countries discussed in this thesis study, Poisson, Negative Binomial, ZIP and ZINB Regression Models, which are the special cases of GLMs class, are used to model the statistical significant relationships between the "child mortality under-5" and the possible risk factors for the lower-middle income countries. These GLMs are compared by using AIC, BIC, AICC as IC and log-likelihood values.