tests/testthat/test-norm_w_to_n.R

In plasso: Cross-Validated (Post-) Lasso

test_that("norm_w_to_n handles non-numeric weights", {
  
  w = c("a", "b", "c")
  
  expect_error(norm_w_to_n(w), "Weights do not have the required format (vector or n x 1 matrix)", fixed=TRUE)
})


test_that("norm_w_to_n handles valid weights (no treatment indicators)", {
  w = c(0.2, 0.3, 0.5)
  result = norm_w_to_n(w)
  
  expect_equal(sum(result), length(w))
})

test_that("norm_w_to_n handles valid weights (with treatment indicators)", {
  w = c(0.2, 0.3, 0.5)
  d = c(1, 0, 1)
  result = norm_w_to_n(w, d)
  
  expect_equal(sum(result[d == 1]), sum(result[d == 0]))
  expect_equal(sum(result), length(w)*2)
})

test_that("norm_w_to_n handles matrix weights (with treatment indicators)", {
  w = matrix(c(1, 1, 1), nrow = 3, ncol = 1)
  d = c(1, 0, 1)
  result = norm_w_to_n(w, d)
  
  expect_equal(sum(result[d == 1]), sum(result[d == 0]))
  expect_equal(sum(result), nrow(w)*2)
  expect_identical(result, as.matrix(c(1.5, 3, 1.5)))
})

test_that("norm_w_to_n handles negative weights - warning", {
  w = c(1,2,-3,4,5)
  w_r =  c(1,2,1,4,5)
  
  expected = w_r / sum(w_r) * length(w)
  
  expect_warning(result <- norm_w_to_n(w), "Negative weights were replaced by the minimum non-negative weight value")
  expect_equal(result, as.matrix(expected))
})