tests/testthat/test_ConditionalScores.R

In kkmann/otsd: Adaptive Optimal Two-Stage Designs in R

context("test sample size")




design <- TwoStageDesign(25, 0, 2, rep(40.0, 5), rep( 2, 5), 5L)
dist   <- Normal()
prior  <- ContinuousPrior(function(x) rep(1/10, length(x)), c(-4, 6))
z1     <- seq(-1, 3, .1)



test_that("conditional sample size maps to actual sample size", {

    css  <- ConditionalSampleSize()
    ess1 <- ExpectedSampleSize(Normal(), prior)
    ess2 <- expected(css, Normal(), prior)

    expect_equal(
        evaluate(css, design, z1),
        n(design, z1),
        tolerance = 1e-4, scale = 1)

    expect_equal(
        evaluate(ess1, design),
        evaluate(ess2, design),
        tolerance = 1e-4, scale = 1)


}) # end 'conditional sample size maps to actual sample size'





context("Test Conditional Power")



test_that("Conditional Power is monotonous", {

    # Define two simple designs
    design1     <- TwoStageDesign(25, 0, 2, rep(40.0, 5), seq(2.0, 0.0, length.out = 5))
    design2     <- TwoStageDesign(25, 0, 2, rep(60.0, 5), seq(2.0, 0.0, length.out = 5))
    dist        <- Normal()
    null        <- PointMassPrior(.0, 1)
    alternative <- PointMassPrior(.4, 1)
    cp          <- ConditionalPower(dist, alternative)

    # Conditional Power is monotonously increasing in n
    expect_gt(
        evaluate(cp, design2, 1),
        evaluate(cp, design1, 1))

    # Conditional Power is monotonously increasing in z1
    expect_gt(
        evaluate(cp, design1, 1.5),
        evaluate(cp, design1, 0.5))

}) # end "Conditional Power is monotonous"



test_that("Conditional power has correct values outside continuation region",{

    design1 <- TwoStageDesign(25, 0, 2, rep(40.0, 5), seq(2.0, 0.0, length.out = 5))
    cp      <- ConditionalPower(Normal(), PointMassPrior(.4, 1))

    expect_equal(
        evaluate(cp, design1, -1),
        0.0,
        tolerance = 1e-4, scale = 1)
    expect_equal(
        evaluate(cp, design1, -1, optimization = TRUE),
        0.0,
        tolerance = 1e-4, scale = 1)

    expect_equal(
        evaluate(cp, design1, 3),
        1.0,
        tolerance = 1e-4, scale = 1)
    expect_equal(
        evaluate(cp, design1, 3, optimization = TRUE),
        1.0,
        tolerance = 1e-4, scale = 1)

}) # end 'Conditional power has correct values outside continuation region'


test_that("Power works",{

    cp     <- ConditionalPower(Normal(), prior)
    pow1   <- Power(Normal(), prior)
    pow2   <- expected(cp, Normal(), prior)

    expect_equal(
        evaluate(pow1, design),
        evaluate(pow2, design),
        tolerance = 1e-4, scale = 1)

}) # end 'Power works'





context("Test class ConditionalScore")



test_that("Conditional scores are vectorized in z1", {

    css    <- ConditionalSampleSize()
    cp     <- ConditionalPower(Normal(), PointMassPrior(.4, 1))

    expect_length(
        evaluate(css, design, z1),
        length(z1)
    )

    expect_length(
        evaluate(cp, design, z1),
        length(z1)
    )

}) # end 'Conditional scores are vectorized in z1'