Prof. Dr. Robert Hable
Faculty of Computer Science
Deggendorf Institute of Technology (DIT)
Technische Hochschule Deggendorf
Dieter-Görlitz-Platz 1, D-94469 Deggendorf
Website at DIT: Prof. Dr. Robert Hable
Email at DIT: robert.hable@th-deg.de
I am a statistician and researcher working in statistical methodology, machine learning, and applied biostatistics. My work combines methodological research with practical applications in medical and scientific projects.
Short profile
I studied mathematics at the University of Bayreuth and received my doctorate in statistics from LMU Munich. I later completed my habilitation at the University of Bayreuth with research on statistical properties of machine learning methods. Since 2019, I have been Professor of Machine Learning at the Deggendorf Institute of Technology.
My research interests lie at the intersection of statistical methodology, machine learning, and applied data analysis. In addition to methodological research, I have worked for many years as a biostatistician in clinical and medical research projects.
I am Associate Editor of the journal Statistics & Probability Letters.
Research Interests
- Applied (bio-)statistics
- Meta-analysis
- Statistical properties of machine learning algorithms
- Robust statistics / imprecise probabilities
Publications
The full list of publications is available on my personal website (German) and on my website at DIT (English).
ORCID profile: 0000-0002-3719-8992
Applied (Bio-)Statistics
Most of my current research activities are in clinical research. This includes statistical support for clinical and medical research projects, ranging from study planning and endpoint definition to data analysis and interpretation. Most of my publications in clinical research are open access and can be found on PubMed.
Meta-analysis
Meta-analysis provides tools for combining evidence from multiple studies. I am interested in the statistical methodology behind meta-analyses as well as in their practical use for deriving reliable overall conclusions from heterogeneous sources of evidence. For an example addressing the reliability of meta-analyses, see:
Statistical Properties of Machine Learning Algorithms
My work has addressed methodological questions concerning statistical properties of machine learning procedures and their use in data analysis. I am particularly interested in the connection between modern machine learning and classical statistical methodology. For example, in the following publication
Hable R.
Asymptotic Normality of Support Vector Machine Variants and Other Regularized Kernel Methods.
Journal of Multivariate Analysis, 106, 92–117, 2012.
I have proven asymptotic normality of Support Vector Machines. You may download a full preprint here.
Robust Statistics / Imprecise Probabilities
Robust statistical methods are designed to provide reliable results even when data contain outliers or when model assumptions are only approximately satisfied. In the publication
Hable R., Ruckdeschel P., Rieder H.
Optimal robust influence functions in semiparametric regression.
Journal of Statistical Planning and Inference, 140(1), 226–245, 2010.
DOI
I investigated a robust version of Cox Regression in survival analysis.
In my Ph.D thesis, I investigated the use of imprecise probabilities in decision theory. Here, you may download the full text of my thesis.
Statistical Consulting
In addition to my academic research, I provide statistical consulting and data analysis services. Further information is available on my German-language website.