tests/testthat/test-bayes-ordinal-preds.R

In bayesMeanScale: Bayesian Post-Estimation on the Mean Scale

set.seed(500)

test_that("make sure all configurations of bayesOrdinalPredsF run without error", {
  
  skip_on_cran()
  skip_if_not_installed('rstanarm')
  skip_if_not_installed('MASS')
  
  housingData <- MASS::housing
  
  rowMiss <- sample(1:nrow(housingData), size=10, replace=F)
  colMiss <- sample(1:ncol(housingData), size=10, replace=T)
  
  for(i in 1:10){
    
    housingData[rowMiss[[i]], colMiss[[i]]] <- NA
    
  }
  
  ordinalModel        <- suppressWarnings(rstanarm::stan_polr(Sat ~ Infl + Type, data=housingData, prior=rstanarm::R2(0.2, 'mean'), refresh=0, iter=200))
  ordinalModelCloglog <- suppressWarnings(rstanarm::stan_polr(Sat ~ Infl + Type, data=housingData, prior=rstanarm::R2(0.2, 'mean'), refresh=0, method='cloglog', iter=200))
  
  expect_no_error(bayesOrdinalPredsF(ordinalModel, at=list(Type=c("Tower", "Apartment")), n_draws=500))
  expect_no_error(bayesOrdinalPredsF(ordinalModel, at=list(Type=c("Tower", "Apartment")), hdi_interval=F, n_draws=500))
  expect_no_error(bayesOrdinalPredsF(ordinalModel, at=list(Type=c("Tower", "Apartment")), at_means=T, n_draws=500))
  expect_no_error(bayesOrdinalPredsF(ordinalModel, at=list(Type=c("Tower", "Apartment")), hdi_interval=F, at_means=T, n_draws=500, data_slice=100))

  expect_no_warning(bayesOrdinalPredsF(ordinalModel, at=list(Type=c("Tower", "Apartment")), n_draws=500))
  expect_no_warning(bayesOrdinalPredsF(ordinalModel, at=list(Type=c("Tower", "Apartment")), hdi_interval=F, n_draws=500))
  expect_no_warning(bayesOrdinalPredsF(ordinalModel, at=list(Type=c("Tower", "Apartment")), at_means=T, n_draws=500))
  expect_no_warning(bayesOrdinalPredsF(ordinalModel, at=list(Type=c("Tower", "Apartment")), hdi_interval=F, at_means=T, n_draws=500, data_slice=100))

logitPreds <- bayesOrdinalPredsF(ordinalModel, at=list(Type=c("Tower")), at_means=T, n_draws=500)$predDraws
logitCheck <- rowSums(subset(logitPreds, select=c(Low, Medium, High)))

cloglogPreds <- bayesOrdinalPredsF(ordinalModelCloglog, at=list(Type=c("Tower")), at_means=F, n_draws=500)$predDraws
cloglogCheck <- rowSums(subset(cloglogPreds, select=c(Low, Medium, High)))
  
  expect_identical(round(min(logitCheck), 6), 1)
  expect_identical(round(max(logitCheck), 6), 1)
  
  expect_identical(round(min(cloglogCheck), 6), 1)
  expect_identical(round(max(cloglogCheck), 6), 1)
  
})