context("Arpr output")

skip_on_cran()

library(laeken)
library(survey)
library(vardpoor)
data(eusilc)
names(eusilc) <- tolower(names(eusilc))
dati = data.frame(IDd = seq(10000 , 10000 + nrow(eusilc) - 1) , eusilc)


SE_lin2 <- function(t, des) {
  variance <-
    survey::svyrecvar(t / des$prob,
                      des$cluster,
                      des$strata,
                      des$fpc,
                      postStrata = des$postStrata)
  sqrt(variance)
}

des_eusilc <-
  svydesign(
    ids = ~ rb030,
    strata =  ~ db040,
    weights = ~ rb050,
    data = eusilc
  )
des_eusilc <- convey_prep(des_eusilc)
des_eusilc_rep <- as.svrepdesign(des_eusilc, type = "bootstrap")
des_eusilc_rep <- convey_prep(des_eusilc_rep)

vardpoor_arprw <-
  linarpr(
    Y = "eqincome",
    id = "IDd",
    weight = "rb050",
    Dom = NULL,
    dataset = dati,
    percentage = 60,
    order_quant = 50L
  )
vardest <- vardpoor_arprw$value
attributes(vardest) <- NULL
vardest <- unlist(vardest)
varse <- SE_lin2 (vardpoor_arprw$lin$lin_arpr, des_eusilc)
attributes(varse) <- NULL
fun_arprw <- svyarpr( ~ eqincome, design = des_eusilc, 0.5, 0.6)
fun_arprw_rep <-
  svyarpr( ~ eqincome, design = des_eusilc_rep, 0.5, 0.6)
convest <- coef(fun_arprw)
attributes(convest) <- NULL
convse <- SE(fun_arprw)
attributes(convse) <- NULL
#domain
vardpoor_arprd <-
  linarpr(
    Y = "eqincome",
    id = "IDd",
    weight = "rb050",
    Dom = "hsize",
    dataset = dati,
    percentage = 60,
    order_quant = 50L
  )
#  point estimates
vardestd <- unlist(vardpoor_arprd$value$arpr)
#  se estimates
varsed <-
  sapply(data.frame(vardpoor_arprd$lin)[, 2:10], function(t)
    SE_lin2(t, des_eusilc))
attributes (varsed) <- NULL
# library convey
fun_arprd <-
  svyby(
    ~ eqincome,
    by = ~ hsize,
    design = des_eusilc,
    FUN = svyarpr,
    quantiles = 0.5,
    percent = 0.6,
    deff = FALSE
  )
convestd <- coef(fun_arprd)
attributes(convestd) <- NULL
convsed <- SE(fun_arprd)

test_that("compare results convey vs vardpoor", {
  expect_equal(vardest, 100 * convest)
  expect_equal(vardestd, 100 * convestd)
  expect_lte(confint(fun_arprw)[1], coef(fun_arprw))
  expect_gte(confint(fun_arprw)[2], coef(fun_arprw))
  expect_equal(coef(fun_arprw), coef(fun_arprw_rep))
  
  # for VARDPOOR replication do not skip:
  skip('density should be computed at the arpt')
  expect_equal(varse, 100 * convse)
  expect_equal(varsed, 100 * convsed)
})