Dependency modeling using copulas with an application of electricity energy in Kurdistan-Iraq

العناوين الأخرى

نمذجة الاعتمادية باستخدام الروابط مع التطبيق في الطاقة الكهربائية في الكردستان-العراق

مقدم أطروحة جامعية

Kiwork, Lusin Immanuil

مشرف أطروحة جامعية

Isa, Sahib Qahraman

أعضاء اللجنة

Ahmad, Nawzad Muhammad Amin
Jalil, Talib S.
Ismail, Muhammad N.
Hasan, Sabah H.
Thanun, Basil Y.

الجامعة

جامعة صلاح الدين

الكلية

كلية الإدارة و الاقتصاد

القسم الأكاديمي

قسم الإحصاء

دولة الجامعة

العراق

الدرجة العلمية

دكتوراه

تاريخ الدرجة العلمية

2012

الملخص الإنجليزي

Copulas approach is useful and important tools for understanding relationships among random variables.

Copulas are powerful and flexible tools for modeling dependence structure of several random variables.

The main aim of this thesis is to study and apply the copula technique to model dependency structures (modeling the joint probabilistic behavior) between electricity generation stations and between the generation and other related resources of stations in the electricity energy in the Kurdistan region.

In this thesis, we study the stochastic dependence structures between daily electricity generations of stations, and between the daily generation and the used fuel of station, based on four power stations : Dokan, Derbandekhan, Erbil and Chamchamal.

The GARCH model is one of the best-known non-linear models for serial dependence and conditional heteroscedastic processes.

Therefore, the copula-GARCH framework is used in this thesis and it generally consists of two components : the first component is to model the marginal distribution of individual series by GARCH (AR-GARCH, ARMA-GARCH and MAGARCH) models with a certain distribution of noise (t and Gaussian distributions are used in our thesis) to filter the linear and the non-linear time dependence in the series of returns, and second is to analyze the dependency through the standardized residuals of these models to estimate a proper copula model (joint distribution) by link the margins together to make capturing the dependence structure best between individual series in each set.

The idea of this thesis is based on the bivariate Archimedean copula and bivariate ellipsoidal copula.

The most proximate estimated copula model (single) is based on the real data in two dimensions by using three estimation methods (the maximum pseudo likelihood), (inversion of Kendall's tau), and (inversion of Spearman's rho).

After a series of goodness of fit tests for copulas, we compare them in each data set by adopting the most reliable criteria for selecting the best fitting copula.

They are the log-maximum likelihood value, the Akaike information criterion, the Bayssian information criterion, and the Cramer-von-Mises statistic.

Based on a comparison of these criteria, the results of the copula modelling show that the best dependence structures (copula models) between the generations and between the generation and other related resources of stations can be modeled by the bivariate (Student-t, product and Clayton) copula provides an adequate representation of the empirical dependence structures.

التخصصات الرئيسية

الرياضيات

الموضوعات

• الطاقة الكهربائية
• النماذج الرياضية
• العراق
• كردستان

عدد الصفحات

165

قائمة المحتويات

• APA
• MLA
• AMA

لغة النص

الإنجليزية

نوع البيانات

رسائل جامعية

رقم السجل

BIM-314692