Originally posted by KrazyHorse
View Post
b.) >> is context specific. A difference of > 50% implies a large deviation from linearity (your objection about irregularities at at the beginning of the distribution aside).
c.) Yeah, 1/N looks about right.
Comment