tests/testthat/test_Error.R

In easyVerification: Ensemble Forecast Verification for Large Data Sets

library(easyVerification)
context("Error scores and skill scores")

fcst <- array(1.5, c(3, 10))
obs <- rep(1, 3)

test_that("Mean error (bias)", {
  expect_is(EnsMe(fcst, obs), "numeric")
  expect_equal(EnsMe(fcst, obs), 0.5)
})

test_that("Mean absolute error", {
  expect_is(EnsMe(fcst, obs), "numeric")
  expect_equal(EnsMae(fcst, obs), 0.5)
  expect_equal(EnsMaess(fcst, fcst, obs), 0)
  expect_equal(EnsMaess(fcst, fcst, obs), EnsErrorss(fcst, fcst, obs, type = "mae"))
})

test_that("Mean squared error", {
  expect_is(EnsMse(fcst, obs), "numeric")
  expect_equal(EnsMse(fcst, obs), 0.5**2)
  expect_equal(EnsMsess(fcst, fcst, obs), 0)
  expect_true(EnsMsess(fcst, fcst, obs) < EnsMsess(fcst, array(0, dim(fcst)), obs))
})

test_that("Root mean squared error", {
  expect_is(EnsRmse(fcst, obs), "numeric")
  expect_equal(EnsRmse(fcst, obs), 0.5)
  expect_equal(EnsRmse(fcst, obs), sqrt(EnsMse(fcst, obs)))
  expect_equal(EnsRmsess(fcst, fcst, obs), 0)
  expect_true(EnsRmsess(fcst, array(0, dim(fcst)), obs) < EnsMsess(fcst, array(0, dim(fcst)), obs))
})