Randomization test/CI on a user-supplied statistic

Runs a randomization test (and, via compute_rand_confidence_ interval(), a randomization confidence interval) for an arbitrary user-supplied statistic, without writing an Inference subclass. The statistic function is called once per permutation as fn(y, w, dead) – plain numeric/integer vectors: the current (possibly permuted) response, treatment assignment, and event indicator. It must return one scalar. For maximum speed, supply custom_randomization_statistic_cpp instead: C++ source defining a function of (NumericVector y, IntegerVector w) or (NumericVector y, IntegerVector w, IntegerVector dead) returning a scalar double, using the same convention as DesignFixedOptimal's custom_objective.

Only a randomization test and randomization confidence interval are available – there is no package point-estimator, Wald path, or bootstrap machinery on this class, so no other action needs disabling.

Super classes

Inference -> InferenceCustomRand -> InferenceRandCustom

Methods

Public methods

• InferenceRandCustom$new()

• InferenceRandCustom$fit()

• InferenceRandCustom$clone()

+ inherited public methods from InferenceCustomRand

• InferenceCustomRand$approximate_randomization_distribution_beta_hat_T()
• InferenceCustomRand$compute_estimate()
• InferenceCustomRand$compute_rand_confidence_interval()
• InferenceCustomRand$compute_rand_two_sided_pval()
• InferenceCustomRand$supports_rand_pval_for_incidence()

+ inherited public methods from Inference

• Inference$capabilities()
• Inference$compute_asymp_confidence_interval()
• Inference$compute_asymp_two_sided_pval()
• Inference$compute_exact_confidence_interval()
• Inference$compute_exact_two_sided_pval_for_treatment_effect()
• Inference$duplicate()
• Inference$get_analysis_data()
• Inference$get_covariates()
• Inference$get_design_object()
• Inference$get_model_formula()
• Inference$get_nonestimable_reason()
• Inference$get_nonestimable_stage()
• Inference$get_optimization_alg()
• Inference$get_response()
• Inference$get_response_type()
• Inference$get_treatment()
• Inference$is_nonestimable()
• Inference$set_optimization_alg()
• Inference$set_seed()
• Inference$supports()

---

InferenceRandCustom$new()

Initialize.

Usage

InferenceRandCustom$new(
  des_obj,
  custom_randomization_statistic_function = NULL,
  custom_randomization_statistic_cpp = NULL,
  verbose = FALSE
)

Arguments

des_obj

See class description.

custom_randomization_statistic_function

See class description.

custom_randomization_statistic_cpp

See class description.

verbose

See class description.

---

InferenceRandCustom$fit()

Calls the user-supplied statistic on the observed data.

Usage

InferenceRandCustom$fit(estimate_only = FALSE)

Arguments

estimate_only

Unused; present for the `fit()` contract.

---

InferenceRandCustom$clone()

The objects of this class are cloneable with this method.

Usage

InferenceRandCustom$clone(deep = FALSE)

Arguments

deep

Whether to make a deep clone.