Λογότυπο αποθετηρίου

Πλοήγηση ανά Συγγραφέα "Tsamtsakiri, Panagiota"

A B C D E F G H I J K L M N O P Q R S T U V W X Y Z
Α Β Γ Δ Ε Ζ Η Θ Ι Κ Λ Μ Ν Ξ Ο Π Ρ Σ Τ Υ Φ Χ Ψ Ω
Φιλτράρετε τα αποτελέσματα πληκτρολογώντας τα πρώτα γράμματα
Τώρα δείχνει 1 - 1 από 1
  • Μικρογραφία εικόνας
    Τεκμήριο
    Modeling multivariate time series
    (2023-11-21) Tsamtsakiri, Panagiota; Τσαμτσακίρη, Παναγιώτα; Athens University of Economics and Business, Department of Statistics; Ntzoufras, Ioannis; Vrontos, Ioannis; Pedeli, Xanthi; Fokianos, Konstantinos; Ombao, Hernando; Barretto-Souza, Wanger; Karlis, Dimitrios
    This thesis deals with the construction of multivariate Integer Generalized Au toregressive Conditional Heteroskedastic(INGARCH) and Conditional Autoregres sive Range(CARR) time series processes. The INGARCH(1,1) model has been alsostudied in multivariate case and implementations in 2 dimensions have been also pre sented recent years. The drawback was found at the restricted values of correlationcoefficient boundaries depending on the way of model’s construction. Based on afamily of copulas a multivariate INGARCH(1,1) model and a CARR(1,1) model arestudied considering interdependencies and self-dependencies respectively. Accordingto model’s complexity, its appropriateness and capability to study data with smallsample sizes are examined and provided with simulations. H-steps ahead forecast ing is considered by taking conditional expectation on volatilities and calculatingmarginal probability mass function.Firstly, considering univariate INGARCH models where volatilities are linearly orlog-linearly expressed offering more flexibility on conditions of stationarity respec tively, a Bayesian Trans-dimensional Markov Chain Monte Carlo is provided. At asecond stage a new alternative family of Sarmanov distribution is also presented inorder to ameliorate boundaries of correlation coefficient and comparison with theknown Sarmanov families are graphically discussed. A multivariate INGARCH(1,1)process is studied based on this alternative Sarmanov distribution and an imple mentation with daily crash counts on three towns in Netherlands is presented. Amultivariate Conditional Aytoregressive Range(CARR(1,1)) model assuming an ex ponential distribution and reconstructing correlation coefficients boundaries is dis cussed. The proposed model is illustrated on bivariate series from the fields ofrenewable sources.