Chi squared (or Χ²) is the name of the probability distrubution that is the sum of the squares of a number of Normal distribution random variables. It is typically encountered in contingency tables, as a measure of whether inbalances in the frequencies represent a real effect or just randomness. This is then used as a metric in some machine learnng techniques, notable decision tree construction, to avoid over learning.
Used in Chap. 21: page 335