gamlj: GAMLj Suite for linear models

library(GAMLj3)
testthat::context("model comparisons")
tol=.001
data("clustermanymodels")
data<-clustermanymodels
data$cluster<-factor(data$cluster)
data$cat2<-factor(data$cat2)
data$cat3<-factor(data$cat3)
data$yord<-factor(data$yord)
data$ybin<-factor(data$ybin)


mod <- GAMLj3::gamlj_lm(
  formula = ycont~x*cat2,
  nested_terms = ~x,
  omnibus = "LRT",
  data = data
)

testthat::test_that("test glm anova comparison option", {
  testthat::expect_equal(mod$main$r2$asDF[3,5],264.805,tolerance = tol)
  testthat::expect_equal(mod$main$r2$asDF[3,4],2,tolerance = tol)
  
})

mod <- GAMLj3::gamlj_lm(
  formula = ycont~x*cat2,
  nested_terms = ~x*cat2,
  omnibus = "LRT",
  data = data
)

testthat::test_that("test glm anova comparison option", {
  testthat::expect_equal(mod$main$r2$asDF[3,5],0,tolerance = tol)
  testthat::expect_equal(mod$main$r2$asDF[3,4],0,tolerance = tol)
  
})

mod <- GAMLj3::gamlj_mixed(
  formula = ycont~x*cat2+(1+x|cluster),
  nested_terms = ~x,
  omnibus = "LRT",
  data = data
)

testthat::test_that("test mixed comparison: fixed", {
  testthat::expect_equal(mod$main$r2$asDF[5,5],279.86,tolerance = tol)
  testthat::expect_equal(mod$main$r2$asDF[5,4],5,tolerance = tol)
  
})

mod <- GAMLj3::gamlj_mixed(
  formula = ycont~x*cat2+(1+x|cluster),
  nested_terms = ~x*cat2,
  nested_re = ~1+(1|cluster),
  omnibus = "LRT",
  data = data
)

testthat::test_that("test mixed comparison: random", {
  testthat::expect_equal(mod$main$r2$asDF[5,5],.458,tolerance = tol)
  testthat::expect_equal(mod$main$r2$asDF[5,4],2,tolerance = tol)
})

mod <- GAMLj3::gamlj_mixed(
  formula = ycont~x*cat2+(1+x|cluster),
  nested_terms = ~x*cat2,
  nested_re = ~1+(1+x|cluster),
  omnibus = "LRT",
  data = data
)

testthat::test_that("test mixed comparison: 0 df", {
  testthat::expect_equal(mod$main$r2$asDF[5,5],0,tolerance = tol)
  testthat::expect_equal(mod$main$r2$asDF[5,4],0,tolerance = tol)
})


mod <- GAMLj3::gamlj_gmixed(
  formula = ybin~x*cat2+(1+x|cluster),
  nested_terms = ~x*cat2,
  nested_re = ~1+(1|cluster),
  data = data
)

testthat::test_that("test mixed comparison: random", {
  testthat::expect_equal(mod$main$r2$asDF[5,5],.142,tolerance = tol)
  testthat::expect_equal(mod$main$r2$asDF[5,4],2,tolerance = tol)
})


mod <- GAMLj3::gamlj_gmixed(
  formula = ybin~x*z+(1+z*x|cluster),
  nested_terms = ~x*z,
  nested_re = ~(1+x+z|cluster),
  data = data
)

mod <- GAMLj3::gamlj_gmixed(
  formula = ybin~x*cat2+(1+x|cluster),
  nested_terms = ~x*cat2,
  nested_re = ~1+(1+x|cluster),
  data = data
)

testthat::test_that("test mixed comparison: 0 df", {
  testthat::expect_equal(mod$main$r2$asDF[5,5],0,tolerance = tol)
  testthat::expect_equal(mod$main$r2$asDF[5,4],0,tolerance = tol)
})


mod <- GAMLj3::gamlj_gmixed(
  formula = ybin~x*cat2+(1|cluster),
  model_type = "probit",
  nested_terms = ~x,
  nested_re = ~1+(1|cluster),
  data = data
)

testthat::test_that("test mixed comparison: probit", {
  testthat::expect_equal(mod$main$r2$asDF[5,5],154.56,tolerance = tol)
  testthat::expect_equal(mod$main$r2$asDF[5,4],2,tolerance = tol)
})

mod <- GAMLj3::gamlj_gmixed(
  formula = yord~x*cat2+(1|cluster),
  model_type = "ordinal",
  nested_terms = ~x,
  nested_re = ~1+(1|cluster),
  data = data
)

testthat::test_that("test mixed comparison: ordinal", {
  testthat::expect_equal(mod$main$r2$asDF[5,5],220.117,tolerance = tol)
  testthat::expect_equal(mod$main$r2$asDF[5,4],2,tolerance = tol)
})


data$ycat<-factor(data$ycat)

suppressWarnings({
mod <- GAMLj3::gamlj_gmixed(
  formula = ycat~x*cat2+(1|cluster),
  model_type = "multinomial",
  nested_terms = ~x,
  nested_re = ~1+(1|cluster),
  data = data
)
})

testthat::test_that("test mixed comparison: ordinal", {
  testthat::expect_equal(mod$main$r2$asDF[2,5],393.26,tolerance = tol)
  testthat::expect_true(is.na(mod$main$r2$asDF[3,5]))
})

mod <- GAMLj3::gamlj_gmixed(
  formula = ypoi~x*cat2+(1|cluster),
  model_type = "poisson",
  nested_terms = ~x,
  nested_re = ~1+(1|cluster),
  data = data
)

testthat::test_that("test mixed comparison: poisson", {
  testthat::expect_equal(mod$main$r2$asDF[5,5],630.53,tolerance = tol)
  testthat::expect_equal(mod$main$r2$asDF[5,4],2)
})


mod <- GAMLj3::gamlj_gmixed(
  formula = ypoi~x*cat2+(1|cluster),
  model_type = "nb",
  nested_terms = ~x,
  nested_re = ~1+(1|cluster),
  data = data
)

testthat::test_that("test mixed comparison: negative binomila", {
  testthat::expect_equal(mod$main$r2$asDF[5,5],183.766,tolerance = tol)
  testthat::expect_equal(mod$main$r2$asDF[5,4],2)
})


##### generalized 

mod <- GAMLj3::gamlj_glm(
  formula = ybin~x*cat2,
  model_type = "logistic",
  nested_terms = ~x,
  data = data
)


testthat::test_that("test gzlm comparison: logistic", {
  testthat::expect_equal(mod$main$r2$asDF[3,5],155.044,tolerance = tol)
  testthat::expect_equal(mod$main$r2$asDF[3,4],2)
})
mod <- GAMLj3::gamlj_glm(
  formula = ybin~x*cat2,
  model_type = "probit",
  nested_terms = ~x,
  data = data
)

testthat::test_that("test gzlm comparison: probit", {
  testthat::expect_equal(mod$main$r2$asDF[3,5],155.493,tolerance = tol)
  testthat::expect_equal(mod$main$r2$asDF[3,4],2)
})

mod <- GAMLj3::gamlj_glm(
  formula = yord~x*cat2,
  model_type = "ordinal",
  nested_terms = ~x,
  data = data
)

testthat::test_that("test gzlm comparison: ordinal", {
  testthat::expect_equal(mod$main$r2$asDF[3,4],207.433,tolerance = tol)
  testthat::expect_equal(mod$main$r2$asDF[3,3],2)
})

mod <- GAMLj3::gamlj_glm(
  formula = yord~x*cat2,
  model_type = "multinomial",
  nested_terms = ~x,
  data = data
)
testthat::test_that("test gzlm comparison: multinomial", {
  testthat::expect_equal(mod$main$r2$asDF[3,4],217.523,tolerance = tol)
  testthat::expect_equal(mod$main$r2$asDF[3,3],8)
})

data$yperc<-data$ycont/(max(data$ycont)+.1)

mod <- GAMLj3::gamlj_glm(
  formula = yperc~x*cat2,
  model_type = "beta",
  nested_terms = ~x,
  data = data
)

testthat::test_that("test gzlm comparison: multinomial", {
  testthat::expect_equal(mod$main$r2$asDF[3,4],259.670,tolerance = tol)
  testthat::expect_equal(mod$main$r2$asDF[3,3],2)
})


mod <- GAMLj3::gamlj_glm(
  formula = ypoi~x*cat2,
  model_type = "nb",
  nested_terms = ~x,
  data = data
)

testthat::test_that("test gzlm comparison: negative binomial", {
  testthat::expect_equal(mod$main$r2$asDF[3,5],168.62,tolerance = tol)
  testthat::expect_equal(mod$main$r2$asDF[3,4],2)
})