This comment is based on “Nassim Taleb” book “The Black Swan” which must read for everyone working with data, statistics and digital twins.
Creating information where is none
Author of the book argues that one of the most useless comments he hears is that some solutions can come from “Robust Statistics.” He (Nassim Taleb) wonders how using these techniques can create information where there is none.
Asymptotic – Idealized situation
What most students of mathematical statistics do (and all products from OEM’s) is assume a structure similar to the close structure games, typically with a priori known probability.
Kurtosis
Kurtosis is how fat the tails of probability curves are, that is, how much rare events play a role. For example, if one event represents 90 percent of 40 years of observations it is usually filtered out as an outlier, so all data set is invalid.
If you miss that one single number, you miss whole thing.
This proves that everything relying on “standard deviation”, “variance”, “least square deviation,” etc. is bogus.
This statistical perspective comes from years of working with real operational data and from learning—sometimes the hard way—how easily numbers can create confidence without understanding. Experience has shown that reality tends to be less impressed by models and abstractions than their creators expect. If you prefer fewer surprises, Nassim Taleb’s books are an efficient way to save time, money, and misplaced certainty.
