TIME SERIES ANALYSIS USING COPULA GAUSS AND AR(1)-N.GARCH(1,1)

*Rezzy Eko Caraka -  Awardee of LPDP Scholarship, Ministry of Finance, Indonesia
Hasbi Yasin -  Departemen Statistika, Fakultas Sains dan Matematika, Universitas Diponegoro, Indonesia
Wawan Sugiarto -  Senior High School 1 Moro, Riau Islands Province, Indonesia
Kadi Mey Ismail -  Awardee of LPDP Scholarship, Ministry of Finance, Indonesia
Published: 30 Jun 2016.
Open Access Copyright (c) 2018 MEDIA STATISTIKA
License URL: http://creativecommons.org/licenses/by-nc-sa/4.0/
Citation Format:
Article Info
Section: Articles
Language: EN
Full Text:
Statistics: 993 2084
Abstract

In this case, the Gaussian Copula is used to connect the data that correlates with the time and with other data sets. Most often, practitioners rely only on the linear correlation to describe the degree of dependence between two or more variables; an approach that can lead to quite misleading conclusions as this measure is only capable of capturing linear relationships. Correlation doesn’t mean causation, prediction using Copula is built on three things that the marginal distribution function, the kernel function, and the function of the Copula. Gaussian Copula involves the covariance matrix are approximated by using kernel functions. Kernel acts as the correlation between the approach of the data values that have the same characteristics. In this case, the characteristics used is the time. The advantage of the kernel function is able to calculate the correlation between random variables that have a realization using data characteristics. The advantage of using the kernel based Copula able to capture the dependencies between data and process data that have the same characteristics with time. Another benefit is that it allows a sequence of random variables have a joint distribution function so that the conditional probability of the prediction can be calculated.

 

Keywords: Binding, Copula, GARCH, Gauss, Time Series

Article Metrics: