Walks a tsr_game (from tsr_play_game()) and computes per-ply evaluation
context using the S3 evaluation core: win probability before/after each move,
the move's delta, the best alternative move, regret, and flags for
missed wins / missed blocks / turning points.
Usage
tsr_analyze_game(game, method = c("heuristic", "rollout"), n = 200L)Arguments
- game
A
tsr_gameobject.- method
One of
"heuristic"(default) or"rollout". Passed to the underlyingtsr_win_prob()/tsr_rate_moves().- n
Integer. Rollouts when
method = "rollout".
Value
A tibble, one row per ply, with columns:
ply, player, cell, i, j, k, l,
win_prob_before, win_prob_after, delta,
best_cell, best_delta, regret,
is_best, missed_win, missed_block, is_turning_point.
Examples
# \donttest{
g <- tsr_play_game("greedy", "random", seed = 2L) # a short, decisive game
tsr_analyze_game(g)
#> # A tibble: 17 × 17
#> ply player cell i j k l win_prob_before win_prob_after
#> <int> <int> <int> <int> <int> <int> <int> <dbl> <dbl>
#> 1 1 1 1 0 0 0 0 0.5 0.533
#> 2 2 2 86 1 1 1 1 0.467 0.5
#> 3 3 1 4 3 0 0 0 0.5 0.550
#> 4 4 2 210 1 0 1 3 0.450 0.463
#> 5 5 1 2 1 0 0 0 0.537 0.731
#> 6 6 2 202 1 2 0 3 0.269 0.279
#> 7 7 1 13 0 3 0 0 0.721 0.772
#> 8 8 2 9 0 2 0 0 0.228 0.255
#> 9 9 1 10 1 2 0 0 0.745 0.878
#> 10 10 2 167 2 1 2 2 0.122 0.145
#> 11 11 1 14 1 3 0 0 0.855 0.937
#> 12 12 2 144 3 3 0 2 0.0634 0.0676
#> 13 13 1 16 3 3 0 0 0.932 0.975
#> 14 14 2 25 0 2 1 0 0.0247 0.0278
#> 15 15 1 8 3 1 0 0 0.972 0.988
#> 16 16 2 104 3 1 2 1 0.0121 0.0129
#> 17 17 1 6 1 1 0 0 0.987 1
#> # ℹ 8 more variables: delta <dbl>, best_cell <int>, best_delta <dbl>,
#> # regret <dbl>, is_best <lgl>, missed_win <lgl>, missed_block <lgl>,
#> # is_turning_point <lgl>
# }