Per-Position Scoring

Returns one row per held-out (sequence, position) with the matched context, predicted probability of the observed next state, and log-likelihood contribution. Useful for diagnostic plots showing where the model is confident vs. surprised.

Usage

score_positions(tree, newdata, worst = NULL)

Arguments

tree

A transitiontrees.

newdata

Sequence data.

worst

Integer or NULL. If given, return only the worst positions — those with the lowest predicted_prob (the moves the model was most surprised by). Default NULL (all positions, in sequence order).

Value

A data.frame with columns sequence_id, position, matched_context, observed, predicted_prob, log_lik.

Examples

fit  <- replicate(40, sample(c("A", "B", "C"), 10, replace = TRUE),
                  simplify = FALSE)
tree <- context_tree(fit, max_depth = 1L)
new  <- replicate(5, sample(c("A", "B", "C"), 10, replace = TRUE),
                  simplify = FALSE)
score_positions(tree, new, worst = 5L)
#>   sequence_id position matched_context observed predicted_prob   log_lik
#> 1           1        4               B        C       0.245614 -1.403994
#> 2           2       10               B        C       0.245614 -1.403994
#> 3           3        2               B        C       0.245614 -1.403994
#> 4           3        6               B        C       0.245614 -1.403994
#> 5           4        2               B        C       0.245614 -1.403994