tests/testthat/test-distribution.R

In campsis: Generic PK/PD Simulation Platform Campsis

library(testthat)

context("Test some of the distributions")

test_that("Normal distribution", {
  set.seed(1)
  dist <- NormalDistribution(mean = 5, sd = 1)
  dist <- dist %>% sample(n = as.integer(100))
  values <- dist@sampled_values
  #hist(values)
  expect_equal(round(mean(values), 1), 5.1)
  expect_equal(round(sd(values), 1), 0.9)
})

test_that("Log-normal distribution", {
  set.seed(1)
  dist <- LogNormalDistribution(meanlog = log(5), sdlog = 0.20)
  dist <- dist %>% sample(n = as.integer(100))
  values <- dist@sampled_values
  #hist(values)
  expect_equal(round(mean(log(values)), 1), 1.6)
  expect_equal(round(sd(log(values)), 1), 0.2)
})

test_that("Uniform distribution", {
  set.seed(1)
  dist <- UniformDistribution(min = 5, max = 25)
  dist <- dist %>% sample(n = as.integer(100))
  values <- dist@sampled_values
  #hist(values)
  expect_equal(round(mean(values), 1), 15.4)
})

test_that("Discrete distribution", {
  set.seed(1)
  dist <- DiscreteDistribution(x = c(1, 2, 3), prob = c(10, 20, 40))
  dist <- dist %>% sample(n = as.integer(100))
  values <- dist@sampled_values
  #hist(values)
  expect_equal(sum(values == 1), 13)
  expect_equal(sum(values == 2), 32)
  expect_equal(sum(values == 3), 55)
})

test_that("Binomial distribution", {
  set.seed(1)
  dist <- BinomialDistribution(trials = 2, prob = 0.5)
  dist <- dist %>% sample(n = as.integer(1000))
  values <- dist@sampled_values
  #hist(values)
  expect_equal(sum(values == 0), 244) # Close to 0.5*0.5*100=25% as expected
  expect_equal(sum(values == 1), 508) # Close to 0.5*100=50% as expected
  expect_equal(sum(values == 2), 248) # Close to 0.5*0.5*100=25% as expected

  dist <- BinomialDistribution(trials = 1, prob = 0.2)
  dist <- dist %>% sample(n = as.integer(1000))
  values <- dist@sampled_values
  #hist(values)
  expect_equal(sum(values == 0), 790) # Close to 0.8*100=80% as expected
  expect_equal(sum(values == 1), 210) # Close to 0.2*100=20% as expected
})