31 August 2012

Are You Crackin' Wise with Me, Palooka?

Among quant types, there's been an on-going battle twixt the frequentists (among whom, Your 'Umble Servent) and the Bayesians. To the unfamiliar: frequentists have all faith and credit in the observed data while Bayesians accept that investigators must have some prior knowledge of the situation, and this knowledge ought not be wasted, but utilized in the analysis.

In the words of the Wiki the point of Bayesian: "...the posterior probability of a random event or an uncertain proposition is the conditional probability that is assigned after the relevant evidence is taken into account."

That "relevant evidence", typically called the prior (again, the Wiki): "A prior is often the purely subjective assessment of an experienced expert."

Since the prior is fuzzy, shall we say, frequentists often use a somewhat derogatory phrase for the process of finding one. So, imagine the chuckle induced when I ran across this turn of phrase from a recent posting: "... a different sampler for sampling from their posteriors." Sometimes, an editor is worth the few ducats they receive.

No comments: