tests/testthat/test-pmi_svd.R

In nlpembeds: Natural Language Processing Embeddings

df_ehr = data.frame(Month = c(1, 1, 1, 2, 2, 3, 3, 4, 4),
                    Patient = c(1, 1, 2, 1, 2, 1, 1, 3, 4),
                    Parent_Code = c('C1', 'C2', 'C2', 'C1', 'C1', 'C1', 'C2',
                                  'C3', 'C4'),
                  Count = 1:9)

spm_cooc = build_df_cooc(df_ehr)

# having codes that sort differently as numerics or chars
df_ehr$Parent_Code = c(1, 10, 10, 1, 1, 1, 10, 2, 3)
spm_cooc_rev = build_df_cooc(df_ehr)

m_pmi = get_pmi(spm_cooc)
m_pmi_rev = get_pmi(spm_cooc_rev)

test_get_pmi = function() {

  sum_cooc = sum(as.matrix(build_spm_cooc_sym(spm_to_df(spm_cooc))))
  expect_equal(m_pmi[1, 1], log(16 * sum_cooc / ((16 + 7) ^ 2)))
  expect_equal(m_pmi[1, 2], log(7 * sum_cooc / ((16 + 7) * (12 + 7))))
  expect_equal(m_pmi[2, 2], log(12 * sum_cooc / ((12 + 7) ^ 2)))

  expect_equal(colnames(m_pmi), paste0('C', 1:4))
  expect_equal(colnames(m_pmi_rev), as.character(c(1, 2, 3, 10)))

  df_pmi = reshape2::melt(m_pmi_rev)
  df_pmi$Var1 %<>% gsub('3', 'C4', .) %>% gsub('2', 'C3', .) %>%
    gsub('10', 'C2', .) %>% gsub('1', 'C1', .) %>% factor
  df_pmi$Var2 %<>% gsub('3', 'C4', .) %>% gsub('2', 'C3', .) %>%
    gsub('10', 'C2', .) %>% gsub('1', 'C1', .) %>% factor
  df_pmi = df_pmi[order(df_pmi$Var2, df_pmi$Var1), ]
  rownames(df_pmi) = seq_len(nrow(df_pmi))

  expect_equal(reshape2::melt(m_pmi), df_pmi)
}
test_that('get_pmi', test_get_pmi())


test_get_svd = function() {

  m_data = scale(iris[1:4])
  m_svd = get_svd(m_data, 2)
  expect_equal(dim(m_svd), c(150, 2))

}
test_that('get_svd', test_get_svd())