Modelling Dependencies in Ultra-High Dimensions

24.01.2016

This project serves to strengthen the cooperation of research groups for statistics and economics and uses digital platforms for knowledge transfer

"Modeling Dependencies in Ultra-High Dimensions" is a project connecting the Ladislaus von Bortkiewicz Chair of Statistics of the Humboldt-Universität zu Berlin with the institutes of Statistics and Mathematics of the University of Prague, Warsaw and Vienna for the purpose of exchanging findings and innovations on the topic, as well as methods of teaching and information distribution.

The project is a continuation of a longstanding successful collaboration between the above mentioned universities on Copula, Applied Quantitative Finance and Multivariate Statistics, which in the past few years has resulted in scientific activities (workshops, seminars and dual learning lectures of statistics), research articles and joint publications which have since been widely used as material for lectures.

Emphasis is placed upon the involvement of PhD-students in research and development. Regular meetings of the professors and PhD-students at annual workshops in the different cities give PhD-students from all partner universities the chance to present and discuss their current work and progress on the topic. They thus benefit from the opportunity to reflect and contribute directly to contemporary research in an environment of dynamic learning.
Furthermore, PhD-students are given the chance to attend relevant conferences in Prague, Warsaw and Vienna.

To strengthen the scope of the collaboration between the research groups involved in Economics and Statistics, professors from partner universities and other international universities are also invited to give talks on the topic by introducing perspectives from departments other than Statistics or Mathematics.

Another special feature of the project is the expansion of the cooperation in a digital field of science with the implementation of two digital platforms into the project: Quantnet and MmStat. The priority of both platforms is given to transparency and reproducibility.
MmStat is an online platform also available on CD, which offers a comprehensive introduction to statistics.
Quantnet is a database-driven online repository of scientific information to freely exchange numerical methods. Codes are written in many programming languages in order to fit the user’s demands and preferences. The aim is to introduce a centralised system that is comprised of documents from different scientific areas, submitted by various authors ranging from professional researchers to university students.

 

Introduction to the topic “Modelling dependencies in ultra-high dimensions”

Modelling dependencies of high-dimensional data plays an important role in Economics and Statistics. The understanding of the interconnections is essential in actuarial science where mathematical and statistical methods are applied to assess risk in insurance, finance and other industries and professions. Our modelling framework considers various dependence structures and improves previous claim reservations based on assumptions of independency. Further research focus is also given to systemic risk, extreme events and copula approach.