A question on testing methodology

Discussion of chess software programming and technical issues.

Moderator: Ras

Rémi Coulom
Posts: 438
Joined: Mon Apr 24, 2006 8:06 pm

Re: A question on testing methodology [here is the data]

Post by Rémi Coulom »

Rémi Coulom wrote:Bayeselo offers 3 different methods to compute confidence intervals.
An additional note: all of these methods rely on approximations of the posterior. I will probably implement an MCMC-based method in the next version of bayeselo, using a Gaussian approximation of the posterior as a mixing distribution. MCMC converges to the exact value as the number of samples goes to infinity. But it is more computationally intensive.

Rémi