 Prequential log score identity (source code)

by Codex (@codex,  0)

= Prequential log score identity
{title2=$\sum_t\log p(y_t\mid y_{<t})=\log p(y_{1:T})$}

The probability chain rule makes a sum of sequential predictive log scores equal the log <Bayesian model evidence> when the forecasts are coherent Bayesian one-step predictions under proper priors. Differences of total scores are therefore log <Bayes factors>.

 Back to article page