Skip to contents

Off-line value of a targeting policy

Source: R/identification.R
proxy_policy_value.Rd

Estimates the expected value of deploying a per-unit targeting policy, \(V(d) = E_X[\,\text{value}\cdot E[Y \mid do(T = d(X)), X] - \text{cost} \cdot d(X)\,]\), from the fitted model alone – no live A/B test. Units that fail the overlap diagnostic are excluded by default and their count is reported, never silently dropped.

Usage

proxy_policy_value(
  model,
  newdata,
  policy,
  value,
  cost = 0,
  t1 = 1,
  t0 = 0,
  exclude_low_overlap = TRUE
)

Arguments

model

An uplift_model.

newdata

A data frame carrying the covariate columns – the population the policy would be deployed on.

policy

A per-unit action specification: a 0/1 vector of length nrow(newdata), a function of the proxy_cate() table returning actions, or one of the strings "all", "none", "optimal".

value

Numeric scalar – the value of one unit of outcome.

cost

Numeric scalar – the cost of treating one unit. Default 0.

t1, t0

The treated and control treatment values. Default 1 and 0.

exclude_low_overlap

Logical – drop overlap-flagged units from the average (and report the count). Default TRUE.

Value

A one-row data.table::data.table with columns policy_value, n_used, n_excluded, n_treated.

See also

proxy_decide(), proxy_overlap()

Other decision: fit_uplift(), gmm_cf_mean(), gmm_cf_tail_prob(), gmm_cf_variance(), gmm_counterfactual(), gmm_intervene(), proxy_cate(), proxy_confounding_gap(), proxy_decide(), proxy_identification_report(), proxy_overlap(), proxy_predict(), proxy_regime_segments(), proxy_retrospective_uplift(), proxy_uplift(), uplift_identification(), uplift_model()

Examples

set.seed(1)
n <- 600L
x <- stats::rnorm(n)
t <- stats::rbinom(n, 1L, 0.5)
y <- 1 + (0.4 + x) * t + stats::rnorm(n, sd = 0.5)
dat <- data.frame(y = y, t = t, x = x)
m <- fit_uplift(dat, "y", "t", "x", N = 2L, regime = "sample",
                max_iter = 80L, seed = 1L)
nd <- data.frame(x = stats::rnorm(200))
proxy_policy_value(m, nd, policy = "optimal", value = 1, cost = 0.3)
#> Excluded 1 low-overlap unit from the policy value.
#>    policy_value n_used n_excluded n_treated
#>           <num>  <int>      <int>     <int>
#> 1:     1.448247    199          1       105