policy {polle}

Policy-class

Description

A function of inherited class "policy" takes a policy data object as input and returns the policy actions for every observation for every (observed) stage.

Details

A policy can either be defined directly by the user using policy_def or a policy can be fitted using policy_learn (or policy_eval). policy_learn returns a policy_object from which the policy can be extracted using get_policy.

Value

data.table with keys id and stage and action variable d.

S3 generics

The following S3 generic functions are available for an object of class policy:

print

Baisc print function

Examples

### Two stages:
d <- sim_two_stage(5e2, seed=1)
pd <- policy_data(d,
                  action = c("A_1", "A_2"),
                  covariates = list(L = c("L_1", "L_2"),
                                    C = c("C_1", "C_2")),
                  utility = c("U_1", "U_2", "U_3"))

# defining a dynamic policy:
p <- policy_def(
  function(L) (L>0)*1,
  reuse = TRUE
)
p
head(p(pd), 5)

# V-restricted (Doubly Robust) Q-learning:
# specifying the learner:
pl <- policy_learn(type = "drql",
                   control = control_drql(qv_models = q_glm(formula = ~ C)))

# fitting the policy (object):
po <- pl(policy_data = pd,
         q_models = q_glm(),
         g_models = g_glm())

p <- get_policy(po)
p

head(p(pd))