rovivor/dsci_553_stat-inf-2

DSCI 553: Statistical Inference and Computation II

Bayesian reasoning for data science. How to formulate and implement inference using the prior-to-posterior paradigm.

Rendered lecture slides: available here

Learning Objectives

By the end of the course, students will be able to:

1. Use Bayesian reasoning when modeling data.
2. Apply Bayesian statistics to regression models.
3. Compare and contrast Bayesian and frequentist methods, and evaluate their relative strengths.
4. Use appropriate statistical libraries and packages for performing Bayesian inference

Lecture Schedule

#	Date	Day	Topic	Reading
1	2019-02-05	Tues	Bayesian tour in the discrete case (some probability recap at same time)
2	2019-02-07	Thurs	More discrete Bayesian analysis
3	2019-02-12	Tues	Debrief from last week's lab. Next: Bayes modelling 101 with PPL, including continuous spaces this time
4	2019-02-14	Thurs	Practical issues, model bestiary
5	2019-02-26	Tues
6	2019-02-28	Thurs
7	2019-03-05	Tues	Looking under the hood of PPLs (if time permits)
8	2019-03-07	Thurs	Computation, continued

Reference Material

• Probabilistic Programming and Bayesian Methods for Hackers (MDS alumni recommended!)
• Introduction to Empirical Bayes: Examples from Baseball Statistics
• Statistical Rethinking: A Bayesian Course with Examples in R and Stan plus exercises converted to PyMC3
• Bayesian Data Analysis book.
• JAGS with R
  • A nice blog post summarizing how to write a JAGS model and input it into R (with the rjags package).
• Quora: For a non-expert, what is the difference between Bayesian and frequentist approaches?