Simple Examples of EB: Baseball
This lecture is based directly on the first few examples in Chapter 16 of SFET: Statistical Foundations for Econometric Techniques. This chapter deals with COUNTING data -- data which counts something or the other and hence is integers. This means Binomial or Poisson Distribution is involved. Some Normal Approximations are also made. Binomials dont have stable variances so the arcsin variance stabilizing distribution is also introduced.
It is important to note that STEIN estimation involves EQUAL VARIANCES, but Empirical Bayes does not. In the book, the two are given equal prominence, but in the video lectures I have omitted the Stein Estimation theory, because I no longer think it is important. I have explained WHY Stein estimators do not matter, and Empirical Bayes is the right method in the video lectures.
BE L21 SFET 16.1 Simple Examples part 1 - Empirical Bayes examples on counting data 1:20m Video