tests/testthat/test-wrap-graph.R

In bnclassify: Learning Discrete Bayesian Network Classifiers from Data

context("Graph")

test_that("Make graph", {
  skip_if_not_installed("graph")
  expect_error(make_graph(nodes = LETTERS[1], from = LETTERS[1], to=1, 
                          weights = 1))
  expect_error(make_graph(nodes = LETTERS[1], from =LETTERS[1],to=LETTERS[1],
                          weights = NULL))
  g <- make_graph(nodes = LETTERS[1:2], from =LETTERS[1],to=LETTERS[2], 
                  weights = 0.1)
  g <- make_graph(nodes = LETTERS[1], from = c(), to = c(), weights = c())
})

test_that("Complete graph", {
  skip_if_not_installed("graph")
  g <- complete_graph(LETTERS[1:5])  
  g <- graph_internal2graph_NEL(g)
  expect_equal(length(graph::edgeNames(g)), 10)
})

test_that("Superimpose node", {
  skip_if_not_installed("graph")
#    Nominal
  e <- list(A='B', B=NULL)
  g <- graph::graphNEL(nodes = LETTERS[1:2], edgeL = e, edgemode = "directed")
  sg <- superimpose_node(graphNEL2_graph_internal(g), 'C')
  sg <- graph_internal2graph_NEL(sg)  
  expect_equal(sort(graph::nodes(sg)), LETTERS[1:3])
  expect_equal(graph::numEdges(sg), 3L)
#    Node already in dag   
  expect_error(superimpose_node(graphNEL2_graph_internal(g), 'A'), 'nodes')
})


test_that("Direct forest", {
  skip_if_not_installed("graph")
  gr <- pairwise_ode_score_contribs(class = 'class', dataset = car, score = 'loglik') 
  af <- max_weight_forest(gr)
  f <- direct_forest(g = af)
  expect_equivalent(graph::ugraph(graph_internal2graph_NEL(f)), graph_internal2graph_NEL(af))
})