Value-at-Risk (VaR) is an important instrument for risk assessment of financial time series. This estimation can be performed in different ways. One possibility is modelling with the aid of GARCH-models and copula functions, which is a very flexible approach. Within this bachelor's thesis the Copula-GARCH method is compared with classic VaR-estimation methods based on the stocks ATX and DAX.
Bachelor's Thesis are written in the course Seminar aus Statistik.
- The formal supervision is held by the lectorer of the seminar.
- The content-related supervisian can be performed by any member of IFAS with a doctoral degree.
Members of the IFAS with Phd. supervise Bachelor and Master theses in Statistics , preferably in the following areas and serve as examiners for the Master’s examination:
Modelling of High-Dimensional Data, Approximative Inference, Biostatistics and Modelling of Genomic Data
Computational Statistics, Cluster Analysis, Mixture Distributions
Experimental Design, Econometrics, Spatial Statistics
Survey Statistics and related topics
Risk Management, Experimental Design, Pension Systems
Time-Series Analysis, Analysis of Longitudinal Data, Survival Analysis, Bayes-Statistics, Medical Statistics
(Generalized) Linear Models, Factor Analysis
Currently the following topics for Bachelor theses are offered for supervision:
• Nonparametric Permutation Tests
|• Effects of Violation of Model Assumptions on Statistical Methods |
• Theory of Probability in Pictures: Visualisation of Important Results of Probability Theory
• Classification and clustering of river sections based on intensity metrics from water level measurements in Austria
|Hainy:||• Simulation Study for the Estimation of Conditional Expectations of Parameters in Nonlinear Mixed Models using Regression Methods|
• The Variance of the Sample Median
• Machine Learning using OfficeCalc
• Counterexamples to the Lindeberg-Lévy theorem
• Randomized Response Techniques for Quatitative Variables
• The Item Count Technique for Sensitive Variables
• Statistical Significance Testing: History, logic of action, and problems
• Corona statistics - statistical methods in the corona pandemic
• Modelling and prediction of the spread of the coronavirus disease 2019
• Financial risk management and pension systems
• Intelligent neural networks
• Time series modelling by neural networks
• Statistical modelling of ecological problems
• Statistics in medicine
• analysis of income of mothers after maternity leave
• Statistical analysis of FMRI data
• Confirmatory Factor Analysis and Structural Equation Modelling
The collection of completed Bachelor's thesis can be found here:
Two Bachelor's Thesis graded as "very good" can be found here (in German)
Magdalena Leitner: Portfolio-VaR-Schätzung mittels bedingter Compula-GARCH Methode
(Portfolio-VaR-Estimation using a conditional Compula-GARCH Method
Daniel Dober: Distanz- und modellbasiertes Clustern eines phänotypischen Datensatzes
(Distance- and modelbased clustern of a phenotypic data set)
This bachelor's thesis conducts a cluster analysis of a phenotypic data set in order to detect class structures with the aid of "unsupervised learning". Therefore, the results of a distance-based cluster analysis, which is generally used for these problems, are compared with those of a model-based cluster analysis. The resulting class structure was used to identify genes having significant effects on the maturity of barley plants.