Normalization

Note: this is the second part of a two-post series where I “fix” some of the problems with crowd-sourced ratings, like those you find for movies or books. (In this series, I look at children’s books.) In the first part, I incorporated a Bayesian prior into the rating calculation to address books with very few ratings sometimes having extreme scores (like 5 out of 5 stars) that likely don’t reflect their actual quality.