Binary logistic log-loss (Geron Eq 4-17, morie.fn grlogc)

Usage

morie_geron_logistic_cross_entropy_cost(X, y, theta, eps = 1e-15)

Arguments

X, y, theta: Design, 0/1 targets and parameters.
eps: Clip guard in (0, 0.5).

Value

List with cost, probabilities, per_instance_loss, n_clipped, accuracy.

Examples

set.seed(5)
X <- cbind(1, rnorm(30))
y <- rbinom(30, 1, plogis(X %*% c(-0.2, 1)))
morie_geron_logistic_cross_entropy_cost(X, y, theta = c(0, 0.5))
#> $cost
#> [1] 0.6681639
#> 
#> $probabilities
#>  [1] 0.3964144 0.6664516 0.3480218 0.5087669 0.7017658 0.4252021 0.4412518
#>  [8] 0.4212399 0.4643389 0.5172567 0.6488106 0.4010986 0.3681419 0.4803184
#> [15] 0.3691465 0.4826337 0.4258859 0.2512450 0.5300658 0.4676259 0.6107001
#> [22] 0.6156050 0.6756783 0.5874371 0.6009691 0.4633805 0.6702453 0.6790451
#> [29] 0.4185957 0.3949869
#> 
#> $per_instance_loss
#>  [1] 0.5048674 0.4057877 1.0554901 0.6757652 1.2098760 0.8551907 0.8181396
#>  [8] 0.5468672 0.7671405 0.6592161 1.0464296 0.5126583 0.4590905 0.7333061
#> [15] 0.4606816 0.7284973 0.5549272 0.2893434 0.6347541 0.6304089 0.9434053
#> [22] 0.9560845 0.3920383 0.5319861 0.5092118 0.6224659 1.1094061 0.3870677
#> [29] 0.5423088 0.5025051
#> 
#> $n_clipped
#> [1] 0
#> 
#> $accuracy
#> [1] 0.6333333
#> 
#> $estimate
#> [1] 0.6681639
#> 
#> $n
#> [1] 30
#>