lppl
v2.0.0
|
double aic | ( | FilterValueType< O, Ts... > & | cache | ) |
Computes the (approximate) Akaike Information Criterion (AIC) of the model.
\(AIC = 2 ( \sum_a dim(r(a)) - \sum_{a\ observed} \log p(r(a)) ) \), where \(r\) is the maximum likelihood record in the cache.
Ts | Types of distributions in the record |
cache | the empirical distribution over records |