Application of Copula Theory in Correlation Analysis and Its Multiple Extension
|School||Tianjin University of Science and Technology|
|Course||Management Science and Engineering|
|Keywords||Copula Multiple correlation structure Vine Related Information|
System first Copula theory and its modeling method, a comprehensive, detailed collate and summarize the graph theory, followed by the introduction of new ideas for the the multiple correlation structure modeling and for multivariate packet decomposition provides a possibility theoretical knowledge into practice, in the four yuan exchange price analysis, not only investigated the correlation structure between the two sequences, also constructed four meta-variable structure model rattan graph theory, graph theory vine The advantage of the method multiple correlation structure modeling. Work and innovation of the main points are as follows: (1) system to collate and summarize Copula Theory and modeling methods. Introduced the concept of Copula nature; conditions Copula general form; the Copula theory leads indicators; Copula modeling parameter estimation problem. (2) the introduction of graph theory vine method for multiple structural modeling to provide new ideas to solve the multi-structural modeling is qualitative issues and the use of the vine multivariate joint distribution of packet decomposition. Copula theory and rattan theory and the iteration structure to take advantage of rattan progressive layers, and eventually build a multivariate structural model and discuss multivariate joint distribution of the feasibility of grouping decomposition using rattan theory given Copula vine modeling parameter estimation method. In addition, the introduction of such concepts as related information, which is used to explain the vine theoretical description of the related structures. (3) combined vine Theory and Copula theory applied to the financial sector. Paper selected four yuan exchange rate data, the first of their data on the number of yield basic statistic analysis, followed by a GARCH model processing, Copula function to consider its twenty-two structure, and the effects of different Copula Function The fitting effect meantime structure. Investigated the correlation structure between the two two sequences, the use of rattan theory, the number yields residual variable multivariate modeling, and were compared with the general Gaussian Copula multivariate modeling, thereafter, Vine Copula model to improve take into account not only the same vine select different Copula function modeling, and also consider the use of different vine structure modeling, and doing the appropriate comparison. The empirical process not only to prove the validity of the multivariate structure model building with vines theory, also by selecting different vine structure in each layer to select different the Copula function of modeling show the flexibility of the structure of the vine.