Variance of policy value estimator via non-contextual adaptive weighting.

Computes the variance of a policy value estimate based on AIPW scores, a policy matrix, and non-contextual adaptive weights.

Usage

aw_var(scores, estimate, policy, evalwts = NULL)

Arguments

scores

Numeric matrix. AIPW scores, shape [A, K]. Must not contain NA values.

estimate

Numeric scalar. Policy value estimate.

policy

Numeric matrix. Policy matrix \(\pi(X_t, w)\), shape [A, K]. Must have the same shape as scores and must not contain NA values.

evalwts

Optional numeric vector. Non-contextual adaptive weights \(h_t\), length A, or NULL.

Value

Numeric scalar. Variance of policy value estimate.

Examples

scores <- matrix(c(0.5, 0.8, 0.6,
                   0.3, 0.9, 0.2,
                   0.5, 0.7, 0.4,
                   0.8, 0.2, 0.6), ncol = 3, byrow = TRUE)
policy <- matrix(c(0.2, 0.3, 0.5,
                   0.6, 0.1, 0.3,
                   0.4, 0.5, 0.1,
                   0.2, 0.7, 0.1), ncol = 3, byrow = TRUE)
estimate <- aw_estimate(scores = scores, policy = policy, evalwts = c(0.5, 1, 0.5, 1.5))
aw_var(scores = scores, estimate = estimate, policy = policy, evalwts = c(0.5, 1, 0.5, 1.5))
#> [1] 0.002994835