tests/testthat/test-hhi.R

local_edition(3)

## concstats_hhi

test_that("concstats_hhi function operates properly", {
  x <- c(0.2, 0.25, 0.4, 0.1, 0.05)
  x1 <- c(0.2, 0.3, 0.25, 0.05, -0.2)
  x1b <- c()
  x2 <- c(0.2, 0.3, 0.4, 0.1, NA)
  x4 <- c(20, 25, 40, 10, 5)
  x5 <- c(0.2, 0.25, 0.4, 0.05, 0.10001)
  x7 <- c(0.2, 0.2499999999, 0.4, 0.1, 0.05)
  x8 <- c(-0.2, -0.3, -0.4, -0.100001)
  x9 <- c(NA, NA, NA, NA, NA)
  xch <- c("a", "b", "c", "d", "e")

  expect_true(any(is.na(x2)), all(!is.na(x2)))

  expect_true(all(round(x) == 0), (abs(x) > 0 & abs(x) <= 1))
  expect_vector(x, ptype = double(), size = 5)
  expect_message(concstats_hhi(x2))

  expect_error(concstats_hhi(xch, !is.numeric(xch)))

  expect_error(concstats_hhi(x9, na.rm = TRUE))
  expect_error(concstats_hhi(x8, na.rm = TRUE))
  expect_error(concstats_hhi(x1b, na.rm = TRUE))
  expect_error(concstats_hhi(x, na.rm = 0))
  expect_error(concstats_hhi(x, normalized = 0))
  expect_error(concstats_hhi(x1, as.logical(any(x < 0))))
  expect_message(concstats_hhi(x2, na.rm = TRUE))
  expect_error(concstats_hhi(anyNA(x2), na.rm = FALSE))
  expect_identical(concstats_hhi(c(1, 2, NA, 4), na.rm = FALSE),
                   NA_real_)
  expect_false(identical(concstats_hhi(c(1, 2, 3, 4),
                                           na.rm = FALSE), NA_real_))

  # digits argument
  expect_error(expect_int(x, digits = c(8, 0)))
  expect_error(expect_int(x, digits = 0))
  # convert to continuous
  act <- concstats_hhi(x)
  exp <- concstats_hhi(x4 / sum(x4))
  expect_equal(act, exp, tolerance = .Machine$double.eps^0.25)
  # Adding trivial noise
  act <- concstats_hhi(x)
  exp <- concstats_hhi(x5)
  expect_equal(act, exp, tolerance = .Machine$double.eps^0.25)
  act <- concstats_hhi(x)
  exp <- concstats_hhi(x7)
  expect_equal(act, exp, tolerance = .Machine$double.eps^0.25)

})

test_that("concstats_hhi returns sum of squared shares as decimal", {
  x3 <- c(0.4, 0.25, 0.2, 0.1, 0.05)
  x4 <- c(20, 25, 40, 10, 5)
  share_2018_hhi <- 0.1234614
  share_2018 <- c(0.012663407, 0.029367501, 0.014456455, 0.012046011,
                  0.007477799, 0.189784408, 0.008738591, 0.015635544,
                  0.012787201, 0.013071539, 0.046268385, 0.006580823,
                  0.009102, 0.00760554, 0.047173998, 0.034356881,
                  0.137813902, 0.016876624, 0.065780114, 0.053775553,
                  0.228519883, 0.030117841)

  act <- concstats_hhi(share_2018)
  exp <- share_2018_hhi
  expect_equal(act, exp, tolerance = .Machine$double.eps^0.25)

  expect_equal(concstats_hhi(share_2018), share_2018_hhi,
               tolerance = .Machine$double.eps^0.25)
  expect_equal(concstats_hhi(x3), sum(x3 ^ 2, na.rm = TRUE))
  expect_equal(concstats_hhi(x4), sum((x4 / sum(x4, na.rm = TRUE))^2))

  expect_true(is.numeric(share_2018_hhi), label = "numeric values returned")

  # {G5.4, G5.4b} Correctness test from hhi package can be checked
  # just visually in the plot
  # (https://joss.theoj.org/papers/10.21105/joss.00828)

  us_2012_hhi <- 0.136192
  us_2012 <- c(1.3, 12.3, 11.5, 9.7, 4.5, 4.2, 0.6, 22.4, 17.9, 7.3, 8.3)
  expect_equal(concstats_hhi(us_2012), us_2012_hhi,
               tolerance = .Machine$double.eps^0.25)
})

test_that("concstats_hhi returns normalized hhi", {

  x <- c(0.2, 0.3, 0.4, 0.1)
  x4 <- c(20, 25, 40, 10, 5)
  share_2018_hhi2 <- 0.08172152
  share_2018 <- c(0.012663407, 0.029367501, 0.014456455, 0.012046011,
                  0.007477799, 0.189784408, 0.008738591, 0.015635544,
                  0.012787201, 0.013071539, 0.046268385, 0.006580823,
                  0.009102, 0.00760554, 0.047173998, 0.034356881,
                  0.137813902, 0.016876624, 0.065780114, 0.053775553,
                  0.228519883, 0.030117841)
  expect_equal(concstats_hhi(share_2018, normalized = TRUE), share_2018_hhi2,
               tolerance = .Machine$double.eps^0.25)
  expect_equal(concstats_hhi(x, normalized = TRUE),
               (sum(x ^ 2) - (1 / sum(x > 0))) / (1 - (1 / sum(x > 0))))
  expect_equal(concstats_hhi(x4, normalized = TRUE),
               as.numeric(sum((x4/sum(x4,na.rm = TRUE))^2) -
                            (1 / sum(x4 > 0))) / (1 - (1 / sum(x4 > 0))))

  checkmate::qexpect(concstats_hhi(x),"N[0,)")
})

# test vector
# loans_2018 <- c(3931577688408.00, 3512006440.00, 1181973244136.00,
# 940212760717.00, 504108367226.00, 553126599129.00, 10349042811855.00,
# 85453595503.00, 112921714388.00, 1080798333076.00, 994455032421.00,
# 1747108722293.00, 868778394702.00, 430724585523.00, 636935285927.00,
# 776207212859.00, 99060073546.00, 110683978608.00, 267147638455.00,
# 4250655893824.00, 920406175495.00, 409127381045.00, 2096325112491.00,
# 760860195590.00, 363222923884.00, 3511443628715.00, 3023084817958.00,
# 601839961663.00, 12092000652237.00, 2036893637665.82)

#' @srrstats {EA6.0, EA6.0e, EA5.8} Values of single-valued objects; for
#'  `numeric` values either using `testthat::expect_equal()` or
#'   equivalent with a defined value for the `tolerance` parameter

# test that sum of x=1
test_that("function throws error when vector outside tolerance", {
  tolerance <- .Machine$double.eps^0.25
  expect_error(concstats_hhi(!isTRUE(all.equal(sum(x, na.rm = TRUE),
                                                    1, tolerance)),
                                  "vector `x` in `concstats_hhi`
                                  does not sum to 1"))

})

Try the concstats package in your browser

Any scripts or data that you put into this service are public.

concstats documentation built on Oct. 12, 2026, 5:07 p.m.