tests/testthat/test_data_strategy.R

In macartan/gbiqq: Make, Update, and Query Binary Causal Models

context(desc = "Testing make_model")

testthat::skip_on_cran()
testthat::test_that(desc = paste("Observational equivalence corresponds to",
                                  "model equivalence."),

  code = {
    m1 <- make_model("X -> Y -> Z <- U")
    m2 <- make_model("U -> Z <- Y <- X")
    expect_equal(m1$dag, m2$dag)
    expect_equal(m1$nodes, m2$nodes)
    expect_equal(m1$nodal_types, m2$nodal_types)
    expect_equal(m1$parameters_df, m2$parameters_df)
    expect_equal(m1$causal_types, m2$causal_types)
  }
)


context(desc = "Testing make_data")

testthat::skip_on_cran()
model <- make_model("X -> M -> Y")

testthat::test_that(desc = "data strategy works",

                    code = {
                      strat <- make_data(model, n = 8)
                      expect_equal(nrow(strat), 8)
                      set.seed(1)
                      strat <- make_data(
                        model,
                        n = 20,
                        nodes = list(c("X", "Y"), "M"),
                        probs = list(1, .5),
                        subsets = list(TRUE, "X==1 & Y==0")
                      )
                      # Test subset functionality
                      subsetM <- strat[!is.na(strat$M), ]
                      expect_equal(subsetM$X, rep(1, nrow(subsetM)))
                      expect_equal(subsetM$Y, rep(0, nrow(subsetM)))


                    })

testthat::test_that(desc = "make_data errors and messages when it should.",

                    code = {
                      model <- make_model("X -> M -> Y")

                      # "If specified, vars, probs, subsets,
                      # should have the same length"
                      expect_error(make_data(
                        model,
                        nodes  = c("X", "Y"),
                        probs = c(1, 2),
                        subsets = c("X==1", "Y==1", "X==0")
                      ))

                      expect_warning(make_data(
                        model,
                        n = 8,
                        n_steps = c(2, 2),
                        nodes = list(c("X", "Y"), "M"),
                        probs = list(1, .5),
                        subsets = list(TRUE, TRUE)
                      ))

                    })