This is an introductory workshop on Bayesian Data Analysis aiming at undergrad students who in the near future may work as a researcher in the academia or industry. I will give an 1-hour overview on the difference between Bayesian and frequentist analysis methods, followed by a 2-hour hands-on session on analysing some simulated data. We will learn how to properly fit a line to data points. We will get to know that we have probably mis-understood some of the statistical concepts. No more reduced-Chi-squared, no more F-test, because a single Bayes Theorem rules them all.
Participants should bring their own laptop and get the following installed before coming to the hands-on session.
Python (better version 2.7.xx)
PyMC3 (the main Bayesian tool we are using)
Jupyter Notebook (iPython interactive interface used together with your web browser)
Pip (tool to download the following Python packages)
- numpy (basic computation)
- scipy (scientific computation)
- matplotlib (for plotting)
- seaborn (for nice contour/corner plots)
You can also choose to install Anaconda for Jupyter following the instruction in the previous Python Training Workshop 2018.
Detailed instruction can be found therein in the links. You are also welcome to contact me (email written below) or Stephen for help.
Jupyter Notebook used in the Workshop
Some recommended readings for Bayesian Data Analysis:
- Data Analysis A Bayesian Tutorial by Sivia & Skilling
- Bayesian Data Analysis by Gelman
- Doing Bayesian Data Analysis by Kruschke