Dr Thomas Süβe (Suesse)
Categorical Data Analysis, Survey Methodology, Analysis of EEG and MEG data, multiple and multivariate tests, mixture distributions, smooth tests, goodness of fit tests, railway logistics, (social) network analysis.
Suesse T. (2010). Analysis and Diagnostics of Categorical Variables with Multiple Outcomes. LAP LAMBERT Academic Publishing. ISBN-13: 978-3-8383-1067-1. ISBN-10: 3838310675. 224 pages. https://www.morebooks.de/store/gb/book/analysis/isbn/978-3-8383-1067-1
Suesse T. and Liu I. (2013). Modelling strategies for repeated multiple response data. International Statistical Review. 81 (2): 230-248.
Suesse T. and Liu I. (2012). Mantel-Haenszel estimators of odds ratios for stratified dependent binomial data. Computational Statistics and Data Analysis. 56 (9): 2705-2717.
Brown B., Suesse T. and Yap V. (2012). Wilson confidence intervals for the two-sample log-odds-ratio in stratified 2 × 2 contingency tables. Communications in Statistics - Theory and Methods. 41 (18): 3355-3370.
Suesse T. (2012). Marginalized exponential random graph models. Journal of Computational and Graphical Statistics. 21 (4): 883-900.
Liu I., Mukherjee B., Suesse T., Sparrow D. and Park S.-K. (2009). Graphical Diagnostics to Check Model Misspecification for the Proportional Odds Model. Statistics in Medicine. 28(3): 412-429.
Liu I. and Suesse T. (2008). The Analysis of Stratified Multiple Responses. Biometrical Journal. 50: 135 - 149.
Haueisen J., Leistritz L., Suesse T., Curio G. and Witte H. (2007). Identifying mutual information transfer in the brain with differential-algebraic modeling: Evidence for fast oscillatory coupling between cortical somatosensory areas 3b and 1. Neuroimage. 37(1): 130-136.
Leistritz L., Putsche P., Schwab K., Hesse W., Suesse T., Haueisen J. and Witte H. (2007). Coupled oscillators for modeling and analysis of EEG/MEG oscillations. Biomedical Engeneering/Biomedizinische Technik. 52 (1): 83-89.
Leistritz L., Suesse T., Haueisen J., Hilgenfeld B. and Witte H. (2006) Methods for parameter identification in oscillatory networks and application to cortical and thalamic 600 Hz activity. Journal of Physiology-Paris. 99 (1): 58-65.
# Hemmelmann C., Horn M., Suesse, T., Vollandt, R. and Weiss S. (2005). New concepts of multiple tests and their use for evaluating high-dimensional EEG data. Journal of Neuroscience Methods. 142: 209-217.
Witte H., Putsche P., Schwab K., Eiselt M., Helbig M. and Suesse T. (2004). On the spatio-temporal organisation of quadratic phase-couplings in ‘tracé alternant’ EEG pattern in full-term newborns. Clinical Neurophysiology. 115: 2308-2315.
# Hemmelmann C., Horn M., Reiterer S., Schack B., Suesse T. and Weiss S. (2004). Multivariate tests for the evaluation of high-dimensional EEG data. Journal of Neuroscience Methods. 139 : 111 - 120.
Note that authors are in alphabetical order on publications marked with #.
Suesse T., Rayner J., and Thas O. (2014). Smooth Tests of Fit for Gaussian Mixtures. Proceedings of European Conference on Data Analysis. 10-12 July 2013 Luxembourg.
Mokhtarian P., Ho T.K., Namazi-Rad M. and Suesse T. (2013). Bayesian Nonparametric Reliability Analysis for a Railway System at Component Level. Proceedings of the IEEE International Conference on Intelligent Transportation Rail Transportation. 30 Aug-1 Sept 2013 Beijing China, 197-202.
Suesse T. (2012). Estimation in autoregressive population models. The Fifth Annual ASEARC Research Conference: Looking to the future, 11-14 Feb 2012, Wollongong NSW: University of Wollongong.
Suesse T. and Liu I. (2008). Diagnostics for Multiple Response Data. Proceedings of 5th International Conference on Probability and Statistics, 5-9 June Smolenice Castle, Slovakia. PROBASTAT 2006, Tatra Mt. Math. Publ. 39 (2008), 105 - 113.
Liu, I., Suesse T., and Mukherjee, B. (2007). Graphical Model-Checking Methods for Proportional Odds Models. The Proceedings of the International Statistical Institute 56th Session. CD-ROM (828.pdf). Lisbon, Portugal.
Suesse T., Haueisen J., Hilgenfeld B., Leistritz L., Witte H. (2004). Oszillatormodelle zur Beschreibung von thalamischer und kortikaler 600 Hz Aktivität (engl.: Oscillator models describing cortical and thalamic 600Hz activity). Proceedings of Biomedical Conference in Ilmenau, Germany. Biomedizinische Technik Suppl. 49: 322 – 323.
Keynote Speaker Presentations
Suesse T. and Chambers, R. (2014). Using Social Network Information in Survey Estimation. “Computational Methods for Survey and Census Data in the Social Sciences” A workshop for statisticians and social scientists. 20-21 June Montreal, Canada.
Suesse T. and Chambers, R. (2013). Using Social Network Information in Survey Estimation. Graybill Conference: Modern Survey Statistics. 9- 12 June Fort Collins, Colorado, USA.
Suesse T. and Liu I. (2011). Modelling Strategies for Repeated Multiple Response Data. NZ Statistics Conference. University of Auckland, Auckland, New Zealand.
Suesse T. and Brown B. (2011). Wilson Confidence Intervals for Stratified 2 by 2 Tables. 4th ASEARC Conference. University at Western Sydney, Sydney, Australia.
STAT231 Probability and Random Variables
STAT904 Statistical Consultancy
PSYC354 Design and Analysis
STAT373 Applied Bayesian Analysis
STAT332 Linear and Generalised Linear Models
Networks, or mathematical graphs, are an important tool for representing relational data, i.e. data on the existence, strength and direction of relationships between interacting actors. Types of actors include individuals, firms and countries. Modeling networks has become more and more important, in particular caused by negative developments in terrorists networks over the past decade, and the currently most widely used class of models are Exponential Random Graph Models (ERGMs). This model approach is useful to explain the underlying generating structure of these data, but is limited in many ways. The PhD project would focus on developing other model approaches that overcome the limitations of ERGMs, for example exploring the use of marginal and transitional models for network data, among others. It also includes theoretical aspects, as consistency of model parameters under non-informative sampling and many more aspects.
Categorical Data Analysis
A common model approach to multivariate binary data is to apply a log-linear model. Log-linear models are useful for describing the joint distribution, but not useful for describing the marginal distribution. A simpler and more effective approach is to apply a generalized linear model (GLM), but it does not account for the dependence of the binary observations. A standard approach that accounts for this dependence is to use generalized estimating equations (GEE). Another less widely known approach is to apply a log-linear model and to constrain the model by a GLM. However current fitting techniques using the iterative proportional fitting (IPF) algorithm are infeasible for large cluster-sizes. The PhD project would focus on the use of Markov-Chain-Monte-Carlo (MCMC) techniques to overcome the limitations of the IPF algorithm. The standard assumption for the model approach is to have equal cluster sizes, the project would also focus on overcoming this limitation, considering smaller cluster sizes as clusters with missing data.
Another related topic would focus on the use of a hybrid method combining generalized mixed models (GLMMs) and marginal models (GLMs). The investigator might be interests in a marginal model that still accounts for some of the variations of model parameters, but not to all. For example in a multi-centre clinical trial, multiple observations might be recorded for each patient and the standard treatment would be compared to a new treatment. Then neither the marginal nor the GLMM approach would be suitable. The PhD project would explore effective model fitting techniques and explore usefulness of such an approach in other applications.
Variance Component Estimation and Testing for Distribution for Mixture Distributions
In (model-based) cluster analysis, mixture distributions are a common tool to model clusters, where each cluster is represented by one multivariate normal distribution, where the mean and variance of that particular multivariate normal characterise important properties of this cluster, as location and scale. Parameter estimation is often achieved by maximum likelihood but resulting in biased variance estimates of the multivariate normal, which might result in incorrect conclusions for cluster analysis. The aim in this project is to obtain unbiased variance estimates. Another issue with model based clustering is checking the validity of the distributional assumptions. This project aims at using smooth tests to check for any distribution of the component densities.
Introduction to Bayesian Data Analysis
See Chapter 9 in Seefeld K. and E. Linder (2007)Statistics Using R with Biological examples, cran.r-project.org/doc/contrib/Seefeld_StatsRBio.pdf
Classification of Two Populations
Three primary variables of interest when examining the relationship between chemical subclinical and overt nonketotic diabetes are glutose intolerance, insulin response to oral glucose and insulin resistance. Using those variables measured on 145 non-obese adults, the question at hand which adults can be considered healthy and sub-clinically diabetic individuals. This project will focus on some popular ‘classification’ methods, as mixture model analysis or discriminant analysis, for discriminant analysis see Chapter 16 in Seefeld K. and E. Linder (2007) Statistics Using R with Biological examples, cran.r-project.org/doc/contrib/Seefeld_StatsRBio.pdf
Benford’s Law – Testing for Manipulated Numbers
Benford's Law, also called the First-Digit Law, refers to the frequency distribution of digits in many real-life sources of data. In this so-called Benford distribution, the number 1 occurs as the leading digit in about 30% of the time, while larger numbers occur in that position less frequently: 9 as the first digit less than 5% of the time. Benford's Law also concerns the expected distribution for digits beyond the first, which approach a uniform distribution. In this project you will work on goodness-of-fit tests, so-called Smooth tests, to test the hypothesis that the numbers follows Benford’s law. If the data don’t follow it, this is often an indication for manipulated or falsified data, so detecting deviations from Benford’s law is an important research avenue.
Models for multivariate Binary Data
The project will investigate model approaches for multivariate binary data, as opposed to univariate independent binary data, and will focus on marginal models, log-linear models and generalised linear mixed models (GLMMs). Popular estimation methods for those models but also interpretation of model parameters are of main interest in this project.
Pre-requisite: STAT332 (Linear&Generalised Linear Models)
Exploring Statistical Methods in Medicine
The project will focus on commonly used statistics in medicine, such as the odds and the risk. For example when comparing the effect of a drug for the treatment of a certain desease, the odds ratio and the relative risk are common statistics in medical sciences to describe the effect and effectiveness of the drug. Suggested reference: Robert H. Riffenburgh (2012) Statistics in Medicine (Chapter 5).