runit-lm.R
In maSAE: Mandallaz' Model-Assisted Small Area Estimators

message("Tests for using lm")
if (interactive()) {
  pkgload::load_all(".")
  source(file.path(
    devtools::as.package(".")["path"], "inst", "runit_tests",
    "setup.R"
  ))
}

fake_weights <- function(df) {
  df[["weights"]] <- 1
  df[["weights"]][df[["x2"]] == 0] <- 0.12
  return(df)
}
suppressWarnings(rm(s1, s2, s0))
data("s1", "s2", "s0", package = "maSAE")
s0$x1 <- s0$x3 <- NULL
s0 <- fake_weights(s0)
s1 <- fake_weights(s1)
s2 <- fake_weights(s2)
s12 <- bind_data(s1, s2)
s012 <- bind_data(s1, s2, s0)
tm <- data.frame(x1 = c(150, 200), x2 = c(23, 23), x3 = c(7, 7.5), g = c("a", "b"))
tm_p <- data.frame(x2 = c(23, 23), g = c("a", "b"))
test_lm_unclustered <- function() {

  # % unclustered

  ## % un-weighted
  ### % two-phase
  #### % partially exhaustive
  object <- maSAE::saObj(data = s12, f = y ~ x1 + x2 + x3 | g, s2 = "phase2", smallAreaMeans = tm_p)
  out <- maSAE::predict(object, use_lm = FALSE)
  outlm <- maSAE::predict(object, use_lm = TRUE)
  RUnit::checkEquals(out, outlm)

  #### % exhaustive
  object <- maSAE::saObj(data = s12, f = y ~ x1 + x2 + x3 | g, s2 = "phase2", smallAreaMeans = tm)
  out <- maSAE::predict(object, use_lm = FALSE)
  outlm <- maSAE::predict(object, use_lm = TRUE)
  RUnit::checkEquals(out, outlm)

  #### % non-exhaustive
  object <- maSAE::saObj(data = s12, f = y ~ x1 + x2 + x3 | g, s2 = "phase2")
  out <- maSAE::predict(object, use_lm = FALSE)
  outlm <- maSAE::predict(object, use_lm = TRUE)
  RUnit::checkEquals(out, outlm)

  ### % three-phase
  object <- maSAE::saObj(data = s012, f = y ~ x1 + x2 + x3 | g, s1 = "phase1", s2 = "phase2")
  out <- maSAE::predict(object, use_lm = FALSE)
  outlm <- maSAE::predict(object, use_lm = TRUE)
  RUnit::checkEquals(out, outlm)

  ## % weighted
  ### % two-phase
  #### % partially exhaustive
  object <- maSAE::saObj(
    data = s12, f = y ~ x1 + x2 + x3 | g,
    s2 = "phase2", smallAreaMeans = tm_p,
    auxiliaryWeights = "weights"
  )
  out <- maSAE::predict(object, use_lm = FALSE)
  outlm <- maSAE::predict(object, use_lm = TRUE)
  RUnit::checkEquals(out, outlm)

  #### % exhaustive
  object <- maSAE::saObj(
    data = s12, f = y ~ x1 + x2 + x3 | g,
    s2 = "phase2", smallAreaMeans = tm,
    auxiliaryWeights = "weights"
  )
  out <- maSAE::predict(object, use_lm = FALSE)
  outlm <- maSAE::predict(object, use_lm = TRUE)
  RUnit::checkEquals(out, outlm)

  #### % non-exhaustive
  object <- maSAE::saObj(
    data = s12, f = y ~ x1 + x2 + x3 | g,
    s2 = "phase2",
    auxiliaryWeights = "weights"
  )
  out <- maSAE::predict(object, use_lm = FALSE)
  outlm <- maSAE::predict(object, use_lm = TRUE)
  RUnit::checkEquals(out, outlm)

  ### % three-phase
  object <- maSAE::saObj(
    data = s012, f = y ~ x1 + x2 + x3 | g,
    s1 = "phase1", s2 = "phase2",
    auxiliaryWeights = "weights"
  )
  out <- maSAE::predict(object, use_lm = FALSE)
  outlm <- maSAE::predict(object, use_lm = TRUE)
  RUnit::checkEquals(out, outlm)
}
if (interactive()) test_lm_unclustered()


test_lm_clustered <- function() {
  # % clustered
  ## % un-weighted
  ### % two-phase
  #### % partially exhaustive
  object <- maSAE::saObj(data = s12, f = y ~ x1 + x2 + x3 | g, s2 = "phase2", smallAreaMeans = tm_p, cluster = "clustid")
  out <- maSAE::predict(object, use_lm = FALSE)
  outlm <- maSAE::predict(object, use_lm = TRUE)
  RUnit::checkEquals(out, outlm)

  #### % exhaustive
  object <- maSAE::saObj(data = s12, f = y ~ x1 + x2 + x3 | g, s2 = "phase2", smallAreaMeans = tm, cluster = "clustid")
  out <- maSAE::predict(object, use_lm = FALSE)
  outlm <- maSAE::predict(object, use_lm = TRUE)
  RUnit::checkEquals(out, outlm)

  #### % non-exhaustive
  object <- maSAE::saObj(data = s12, f = y ~ x1 + x2 + x3 | g, s2 = "phase2", cluster = "clustid")
  out <- maSAE::predict(object, use_lm = FALSE)
  outlm <- maSAE::predict(object, use_lm = TRUE)
  RUnit::checkEquals(out, outlm)

  ### % three-phase
  object <- maSAE::saObj(data = s012, f = y ~ x1 + x2 + x3 | g, s1 = "phase1", s2 = "phase2", cluster = "clustid")
  out <- maSAE::predict(object, use_lm = FALSE)
  outlm <- maSAE::predict(object, use_lm = TRUE)
  RUnit::checkEquals(out, outlm)


  ## % weighted
  ### % two-phase
  #### % partially exhaustive
  object <- maSAE::saObj(
    data = s12, f = y ~ x1 + x2 + x3 | g, s2 = "phase2", smallAreaMeans = tm_p, cluster = "clustid",
    auxiliaryWeights = "weights"
  )
  out <- maSAE::predict(object, use_lm = FALSE)
  outlm <- maSAE::predict(object, use_lm = TRUE)
  RUnit::checkEquals(out, outlm)

  #### % exhaustive
  object <- maSAE::saObj(
    data = s12, f = y ~ x1 + x2 + x3 | g, s2 = "phase2", smallAreaMeans = tm, cluster = "clustid",
    auxiliaryWeights = "weights"
  )
  out <- maSAE::predict(object, use_lm = FALSE)
  outlm <- maSAE::predict(object, use_lm = TRUE)
  RUnit::checkEquals(out, outlm)

  #### % non-exhaustive
  object <- maSAE::saObj(
    data = s12, f = y ~ x1 + x2 + x3 | g, s2 = "phase2", cluster = "clustid",
    auxiliaryWeights = "weights"
  )
  out <- maSAE::predict(object, use_lm = FALSE)
  outlm <- maSAE::predict(object, use_lm = TRUE)
  RUnit::checkEquals(out, outlm)

  ### % three-phase
  object <- maSAE::saObj(
    data = s012, f = y ~ x1 + x2 + x3 | g, s1 = "phase1", s2 = "phase2", cluster = "clustid",
    auxiliaryWeights = "weights"
  )
  out <- maSAE::predict(object, use_lm = FALSE)
  outlm <- maSAE::predict(object, use_lm = TRUE)
  RUnit::checkEquals(out, outlm)
}
if (interactive()) test_lm_clustered()