Il ruolo del quozíente di rilevanza nelle inferenze lambda-predittive

Abstract

This study deals with the probabilistic framework proposed by Johnson e Carnap, which leads to lambda-predictive inference, where prior probabilities are to be modified, after observing a sample evidence, following an additive pattern focused on a parameter, usually denoted by lambda, which represents the weight of prior information. This procedure is suitable only when an important condition holds: the invariance of the relevance quotient, which is fundamental in this context; therefore, the meaning and the use of such quotient in lambda-predictive inference are thoroughly developed. Finally, a sequential sample technique is proposed, based on the value of the relevance quotient.