No abstract

No abstract

Objective: Multiple pregnancies occurred more frequently with being widespread of the assisted reproduction techniques. The recent researches showed that the possibility of multiple pregnancies has been increased by some factors such as twin pregnancy experience in the family, mothers at later ages, social properties and the number of live-born infants. The main aim of this study is to identify the statistically significant factors affecting the multiple pregnancies using count data models. Material and Methods: In this study, the number of babies born alive for the pregnant who have multiple pregnancy diagnose are considered in 2015 for a specific location, Ankara province in Turkey. For this purpose, the effects of mother's age, the number of pregnancy, the method of delivery, mother's blood type and previous births of the mother are statistically analyzed using various count data models. Results: Quasi Poisson and Conway-Maxwell-Poisson (COM) regression models are used due to under-dispersion problem and these models are compared for the data set. As a result of comparison, the advantage of COM Poisson regression to other count regression models are illustrated with a real data set. According to results of COM Poisson, the number of pregnant and method of delivery (specifically cesarean type) has significant impact on the number of live-born infants at %99 significance level and the blood type of pregnant has significant impact at %95 significance level. However, the cesarean delivery has negative impact on the number of live-born infants. Conclusion: The main indicator of existence of under-dispersion is significant so fitting COMPoisson to the data in this study is meaningful. For the model selection, based on AIC values of Poisson and COM-Poisson models, the latter one is smaller hence fits better. After recovering the problem of underdispersion for the count data with COM Poisson regression, the number of pregnancy and the method of delivery has been determined as the best predictor. Besides, the blood types were identified as additional explanatory variable, but with lower significance level.Keywords: Multiple Pregnancy; quasi poisson regression; under-dispersion, COM Poisson regression ÖZET Amaç: Günümüzde yardımcı üreme tekniklerinin giderek yaygınlaşması ile çoğul gebeliklere daha sık rastlanmaya başlanmıştır. Son yıllarda yapılan araştırmalara göre, ailede daha önceden ikiz gebelik olması, ileri anne yaşı, toplumsal özellikler ve canlı doğan bebek sayısı gibi faktörlerin çoklu gebelik görülme olasılığını artırdığı gözlemlenmiştir. Bu çalışmanın esas amacı çoğul gebelikleri etkileyen istatistiksel olarak anlamlı faktörleri sayım modelleri kullanarak belirlemektir. Gereç ve Yöntemler: Bu çalışmada 2015 yılında Ankara ilinde çoklu gebelik tanısı konmuş gebelerin doğum sonuçlarına göre canlı doğan bebek sayısı dikkate alınmıştır. Bu amaçla, annenin yaşı, kaçıncı gebeliği olduğu, doğum yöntemi, kan grubu ve önceki doğum durumu gibi faktörlerin etkileri istatistiksel olarak sayım modelleri kullanı...

The increase of the product variety in the financial markets requires a clear understanding of the dependence between such instruments for the decision-makers. For a few decades, such dependence structures were often modeled with symmetric copula families. However, financial data may reveal an asymmetric structure, which can be determined via directional dependence measures in the context of copulas. Previously, some asymmetric copula models were proposed in different ways using Khoudraji’s device. But they are merely used for financial time series data in a broader sense. In this study, a new set of asymmetric copulas were defined by using one parameter of Archimedean copula families. For this aim, widely used copula families were studied and the corresponding directional dependence measures were analyzed. To illustrate the efficiency of the parameter estimation method, a small simulation scenario consisting of an asymmetric dependence pattern was carried out. Thereafter, the proposed asymmetric bi-variate copulas with directional dependence coefficients were investigated for two different stock market data. The study’s primary findings suggested that the newly generated asymmetric models might be useful for directional dependence. Especially, the estimated directional dependence coefficients can serve as an indicator to explain the variability of one stock in terms of the other.

No abstract