tests/testthat/test-plot.R

In distr6: The Complete R6 Probability Distributions Interface

test_that("missing d/p/q", {
  expect_error(expect_message(
    plot(Distribution$new("s", pdf = function(x) x, type = Reals$new()), fun = "cdf"),
    "does not have a cdf expression"
  ), "No plottable")
  expect_error(expect_message(
    plot(Distribution$new("s", cdf = function(x) x, type = Reals$new()), fun = "pdf"),
    "does not have a pdf expression"
  ), "No plottable")
  expect_error(expect_message(
    plot(Distribution$new("s", pdf = function(x) x, type = Reals$new()), fun = "quantile"),
    "does not have a quantile expression"
  ), "No plottable")
})

test_that("errors", {
  expect_error(plot(Binomial$new(), fun = "los"), "Function unrecognised")
})

test_that("silent", {
  expect_silent(plot(Binomial$new(), "all"))
  expect_silent(plot(Binomial$new()))
  expect_silent(plot(Binomial$new(), "pdf"))
  expect_silent(plot(Binomial$new(), c("pdf", "cdf", "quantile")))
  expect_silent(plot(Binomial$new(), plot = FALSE))
  expect_silent(plot(Binomial$new(), arrange = FALSE))
  expect_silent(plot(Normal$new(), "all"))
  expect_silent(plot(Normal$new()))
  expect_silent(plot(Normal$new(), "pdf"))
  expect_silent(plot(Normal$new(), plot = FALSE))
  expect_silent(plot(Normal$new(), arrange = FALSE))
  expect_silent(plot(Geometric$new(), "all"))
})

test_that("structure", {
  expect_equal(
    plot(Binomial$new(), "cdf", npoints = 1, plot = FALSE)[, 1:2],
    data.table::data.table(points = 0:10, cdf = pbinom(0:10, 10, 0.5))
  )
})

test_that("pars", {
  expect_silent(plot(Geometric$new(),
    fun = "all", ask = TRUE, pdf_col = 1, cdf_col = 2,
    quantile_col = 3, hazard_col = 4, cumhazard_col = 5, survival_col = 6
  ))
  expect_silent(plot(Normal$new(),
    fun = "all", ask = TRUE, pdf_col = 1, cdf_col = 2,
    quantile_col = 3, hazard_col = 4, cumhazard_col = 5, survival_col = 6
  ))
})

test_that("use_rand", {
  expect_silent(plot(Wald$new(), fun = "cdf"))
})

dexpo <- function(x) {
  m1 <- self$getParameterValue("rate")
  m2 <- exp(-1 * self$getParameterValue("rate") * x)
  return(m1 * m2)
}
cexpo <- function(x) {
  m1 <- exp(-1 * self$getParameterValue("rate") * x)
  return(1 - m1)
}

ps <- getParameterSet.Exponential()

continuousTester <- Distribution$new("Continuous Test", "ContTest",
  support = PosReals$new(),
  symmetric = TRUE, type = PosReals$new(zero = T),
  pdf = dexpo,
  cdf = cexpo,
  parameters = ps
)
test_that("use_support", {
  expect_message(plot(continuousTester, xlim = c(0, 10)), "No quantile or")
})

test_that("multivariate", {
  skip_if_not_installed("plotly")
  dist <- MultivariateNormal$new()
  expect_silent(plot(dist, fun = "pdf", npoints = 10))
  expect_error(expect_message(plot(dist, fun = "cdf", npoints = 10)), "No plottable")
  dist <- Multinomial$new()
  expect_silent(plot(dist, fun = "pdf", npoints = 10))
  expect_error(expect_message(plot(dist, fun = "cdf", npoints = 10)), "No plottable")
  dist <- EmpiricalMV$new(data.frame(1:10, 1:10))
  expect_silent(plot(dist, fun = "pdf", npoints = 10))
  expect_silent(plot(dist, fun = "cdf", npoints = 10))
  expect_error(plot(EmpiricalMV$new(matrix(1, ncol = 3))), "over two variables")
})

test_that("matdist", {
  set.seed(1)
  pdf <- runif(200)
  mat <- matrix(pdf, 20, 10)
  mat <- t(apply(mat, 1, function(x) x / sum(x)))
  colnames(mat) <- 1:10
  d <- as.Distribution(mat, fun = "pdf")
  expect_silent(plot(d, "pdf", xlab = "x", ylab = "p(x)"))
  expect_silent(plot(d, "cdf", xlab = "x", ylab = "F(x)"))
  expect_silent(plot(d, "survival", xlab = "x", ylab = "S(x)"))
  expect_silent(plot(d, "hazard", xlab = "x", ylab = "h(x)"))
  expect_silent(plot(d, "cumhazard", xlab = "x", ylab = "H(x)"))
})