WinBUGS and R
During the last years, Bayesian statistical modelling has become one of the most fashionable statistical approaches in scientific and technological applications. There are at least two reasons for this trend. One is the current demand of building statistical models which deal with multiple sources of variability. Bayesian models are well suited for this task and they provide an avenue to combine complex information in a coherent form. Successful examples of this approach include innumerable applications of hierarchical modelling.
The other reason is the computation revolution produced by the rediscovery of Markov chain Monte Carlo (MCMC) techniques in statistics, together with their implementation in public domain and friendly to use statistical software like WinBUGS and many R packages. As a result, researchers can construct arbitrary complex statistical models, which may better reflect the phenomena of interest.
This course has two aims. First, it presents a conceptual introduction to Bayesian statistical techniques to practitioners and researchers. Second, it provides a large number of case studies analysed with R and WinBUGS. These examples can serve as a ready to use templates for immediate applications. The course follows a practical perspective rather than a theoretical one. We focus on model building with WinBUGS and interacting WinBUGS with R. Some advantages of running WinBUGS within R include the ability to perform effective model checking, sensitivity and convergence analysis and using the extremely powerful graphical capabilities of R. The target audience are statisticians and data analysts who have familiarity in classical methods such as generalized linear models and random-effects modelling. Neither experience in Bayesian methods nor in WinBUGS will be assumed. However, some working experience in R will be.
The course's working language and documentation is english; discussion and support will be available in german, english and spanish.This course is very popular with students and post-graduates. Therefore, we are offering a special reduced price for students..
Trainer und Dozenten
Dr. rer. nat. Pablo E. Verde is working in the range of statistical modelling in medical and clinical research, currently in synthesis of evidence (meta-analyses). He chairs the Biometrics Task Group in the Coordination Centre for Clinical Studies at Heinrich-Heine-Universität Düsseldorf and conducts research at the university's Institute for Medical Sociology. He has more than 20 years of international experience in statistical consultancy, research and teaching in the domains of medical science, agriculture, health research and risk analysis/financial econometrics.
Pablo is an expert on statistical software R and WinBUGS for MCMC calculations. Since 1998, he is an active member of the R community, being in charge of translation of R into spanish language.
Since 1990, he is teaching the application of S and R, at first for the financial sector and since 2000 in the academic realm.
Pablo is Visiting Lecturer of the Department of Statistics at Stanford University and a Stanford Community Member since 2007. In 2000, Pablo became a member of the Royal Statistical Society.
This course addresses data analysts with practical experience in the R programming language and software environment for statistical computing and graphics.
Introduction to modern Bayesian inference
- Introduction to different kinds of probabilities
- Monte Carlo simulation with WinBUGS and R
- Modern statistical Bayesian inference
- Why do MCMC methods work ?
Introduction to modern Bayesian inferenceFortsetzung vom Vormittag
Connecting R with WinBUGS
- The role of prior distributions in Bayesian inference
- Running WinBUGS with R: examples of Bayesian regression models
- Sensitivity analysis, model criticisms and comparison
- Modelling binary data
- Modelling count data
Connecting R with WinBUGSFortsetzung vom Vormittag
Introduction to Hierarchical Modelling
- Modelling multiple sources of variability
- Examples in hierarchical modelling
(we encourage participants to analyse their own data)
Wer möchte, reist bis 22 Uhr am Vortag an und nutzt den Abend bereits zum Fachsimpeln am Kamin oder im Park.
An den Kurstagen dann von 9-18 Uhr (mit 2 Kaffee- und 1 Mittagspause) etwa 60% Schulungen und 40% Übungen. Selbstverständlich arbeitet jeder Teilnehmer am von uns gestellten Notebook oft parallel zum Referenten mit.
Anschließend Abendessen und Angebote für Fachsimpeln, Ausflüge uvm. Wir schaffen eine Atmosphäre, in der Fachleute sich ungezwungen austauschen. Wer das nicht will, wird zu nichts gezwungen und findet auch jederzeit Ruhe.
Preis und Dauer
961,00 € + 19% MwSt. = 1.143,59 €
Termine und Anmeldung
- Es steht noch kein Termin für diesen Kurs fest.