tests/testthat/test-kernelweights-direct.R

In np: Nonparametric Kernel Smoothing Methods for Mixed Data Types

test_that("direct kernel-weight helper matches npksum.default with evaluation data", {
  kw.fun <- getFromNamespace(".np_kernel_weights_direct", "np")
  ksum.fun <- getFromNamespace("npksum.default", "np")
  set.seed(4021)
  n <- 60
  x <- runif(n)
  z <- rnorm(n)
  tx <- data.frame(x = x, z = z)
  ex <- data.frame(x = seq(min(x), max(x), length.out = 21), z = seq(min(z), max(z), length.out = 21))
  y <- sin(x) + 0.1 * z + rnorm(n, sd = 0.05)

  bw <- npregbw(
    xdat = tx,
    ydat = y,
    regtype = "ll",
    bws = c(0.3, 0.4),
    bandwidth.compute = FALSE
  )

  kw.direct <- kw.fun(
    bws = bw,
    txdat = tx,
    exdat = ex,
    leave.one.out = FALSE,
    bandwidth.divide = TRUE
  )

  kw.ref <- ksum.fun(
    bws = bw,
    txdat = tx,
    exdat = ex,
    return.kernel.weights = TRUE,
    bandwidth.divide = TRUE
  )$kw

  expect_equal(kw.direct, kw.ref, tolerance = 1e-12)
})

test_that("direct kernel-weight helper matches npksum.default leave-one-out", {
  kw.fun <- getFromNamespace(".np_kernel_weights_direct", "np")
  ksum.fun <- getFromNamespace("npksum.default", "np")
  set.seed(4022)
  n <- 50
  x <- runif(n)
  y <- cos(2 * pi * x) + rnorm(n, sd = 0.04)
  tx <- data.frame(x = x)

  bw <- npregbw(
    xdat = tx,
    ydat = y,
    regtype = "lc",
    bws = 0.25,
    bandwidth.compute = FALSE
  )

  kw.direct <- kw.fun(
    bws = bw,
    txdat = tx,
    exdat = NULL,
    leave.one.out = TRUE,
    bandwidth.divide = TRUE
  )

  kw.ref <- ksum.fun(
    bws = bw,
    txdat = tx,
    leave.one.out = TRUE,
    return.kernel.weights = TRUE,
    bandwidth.divide = TRUE
  )$kw
  if (nrow(kw.ref) == ncol(kw.ref))
    diag(kw.ref) <- 0

  expect_equal(kw.direct, kw.ref, tolerance = 1e-12)
})