THE FUTURE OF STATISTICS
Bayesian inference
Training description
The training covers a broad range of topics on Bayesian inference and its implementation in R.
Duration: 2 days 8 hours each (including an hour lunch break)
Requirements: basic knowledge of statistics.
Training agenda
Part one: The foundation
- Crisis in statistics – why Bayesian approach is the future?
- Why p-value still rules at universities?
- Issues and tools of the classic approach
- The alternative of Bayesian approach and its workflow
Part two: The theory of everything
- Prior function, probability function and posterior function
- Selecting the optimal prior function
- Conjunction in Bayesian approach
- Verification of hypotheses in Bayesian approach – Bayes factor
- M type error and S type error
- Estimating future values using posteriori function
Part three: Let’s code!
- Probabilistic programming languages (PPL) – the story
- Overview of languages PPL – BUGS, Jags and Stan
- MCMC – Bayesian sampling approach
Part four: Checking won’t hurt
- Diagnosing of Bayesian models
- Detection of MCMC sampling abnormalities
Part five: Specialised-haul
- Overview of R/Python packages supporting modelling in Bayesian approach
Part six: The small ones matter, too
- Shrinkage – the concept of shrinking
- Hierarchical modelling in Bayesian approach
Contact us about closed training