tests/testthat/test_randomForest.R

In ModelOriented/randomForestExplainer: Explaining and Visualizing Random Forests in Terms of Variable Importance

library(randomForest)
library(dplyr)
load(system.file("testdata/test_randomForest.rda", package="randomForestExplainer", mustWork=TRUE))
# Test input generated by:
# library(randomForest)
# set.seed(12345)
# rf_c <- randomForest(Species ~ ., data = iris, localImp = TRUE, ntree = 2)
# rf_r <- randomForest(mpg ~ ., data = mtcars, localImp = TRUE, ntree = 2)
# rf_u <- randomForest(x = iris, keep.forest = TRUE, localImp = TRUE, ntree = 2)
# save(rf_c, rf_r, rf_u, file = "inst/testdata/test_randomForest.rda")


# Test randomForest classification forests -------------------

test_that("measure_importance works for randomForest classification forests", {
  imp_df <- measure_importance(rf_c, mean_sample = "all_trees",
                               measures = c("mean_min_depth", "accuracy_decrease", "gini_decrease",
                                            "no_of_nodes", "times_a_root", "p_value"))
  expect_equal(as.character(imp_df$variable),
               c("Petal.Length", "Petal.Width", "Sepal.Length", "Sepal.Width"))
})

test_that("important_variables works for randomForest classification forests", {
  imp_vars <- important_variables(rf_c, k = 3,
                                  measures = c("mean_min_depth", "accuracy_decrease", "gini_decrease",
                                               "no_of_nodes", "times_a_root", "p_value"))
  expect_equal(imp_vars, c("Petal.Width", "Petal.Length", "Sepal.Length"))
})

test_that("min_depth_distribution works for randomForest classification forests", {
  min_depth_dist <- min_depth_distribution(rf_c)
  expect_equal(min_depth_dist[min_depth_dist$tree == 1 & min_depth_dist$variable == "Petal.Width", ]$minimal_depth, 0)
})

test_that("min_depth_interactions works for randomForest classification forests", {
  # depth_int returns matrix
  min_depth_int <- min_depth_interactions(rf_c, vars = c("Petal.Width"))
  expect_equal(
    min_depth_int[min_depth_int$interaction == "Petal.Width:Sepal.Length", ]$mean_min_depth,
    0,
    ignore_attr = "dim"
    )
})

test_that("plot_multi_way_importance works for randomForest classification forests", {
  p <- plot_multi_way_importance(rf_c)
  expect_silent(print(p))
})

test_that("plot_min_depth_distribution works for randomForest classification forests", {
  p <- plot_min_depth_distribution(rf_c)
  expect_silent(print(p))
})

test_that("plot_min_depth_interacxtions works for randomForest classification forests", {
  int_frame <- min_depth_interactions(rf_c)
  expect_equal(dim(int_frame), c(16, 6))
  p <- plot_min_depth_interactions(rf_c)
  expect_silent(print(p))
})

test_that("plot_predict_interaction works for randomForest classification forests", {
  p <- plot_predict_interaction(rf_c, data = iris, variable1 = "Petal.Width", variable2 = "Petal.Length")
  expect_silent(print(p))
})

test_that("plot_importance_ggpairs works for randomForest classification forests", {
  p <- plot_importance_ggpairs(rf_c)
  expect_silent(suppressMessages(print(p)))
})

test_that("plot_importance_rankings works for randomForest classification forests", {
  p <- plot_importance_rankings(rf_c)
  expect_silent(suppressWarnings(suppressMessages(print(p))))
})

# Test randomForest regression forests ---------------

test_that("measure_importance works for randomForest regression forests ", {
  imp_df <- measure_importance(rf_r, mean_sample = "all_trees",
                               measures = c("mean_min_depth", "mse_increase", "node_purity_increase",
                                            "no_of_nodes", "times_a_root", "p_value"))
  expect_equal(as.character(imp_df$variable),
               c("am", "carb", "cyl", "disp", "drat", "gear", "hp", "qsec", "vs", "wt"))
})

test_that("important_variables works for randomForest regression forests ", {
  imp_vars <- important_variables(rf_r, k = 3,
                                  measures = c("mean_min_depth", "mse_increase", "node_purity_increase",
                                               "no_of_nodes", "times_a_root", "p_value"))
  expect_equal(imp_vars, c("cyl", "disp", "hp", "wt"))
})

test_that("min_depth_distribution works for randomForest regression forests ", {
  min_depth_dist <- min_depth_distribution(rf_r)
  expect_equal(
    min_depth_dist[min_depth_dist$tree == 1 & min_depth_dist$variable == "cyl", ]$minimal_depth,
    0)
})

test_that("min_depth_interactions works for randomForest regression forests ", {
  min_depth_int <- min_depth_interactions(rf_r, vars = c("cyl"))
  # returns matrix
  expect_equal(
    min_depth_int[min_depth_int$interaction == "cyl:wt", ]$mean_min_depth,
    1,
    ignore_attr = "dim")
})

test_that("plot_multi_way_importance works for randomForest regression forests ", {
  p <- plot_multi_way_importance(rf_r)
  expect_silent(suppressWarnings(print(p)))
})

test_that("plot_min_depth_distribution works for randomForest regression forests ", {
  p <- plot_min_depth_distribution(rf_r)
  expect_silent(print(p))
})

test_that("plot_min_depth_interactions works for randomForest regression forests ", {
  p <- plot_min_depth_interactions(rf_r)
  expect_silent(print(p))
})

test_that("plot_predict_interaction works for randomForest regression forests ", {
  p <- plot_predict_interaction(rf_r, data = mtcars, variable1 = "cyl", variable2 = "wt")
  expect_silent(print(p))
})

test_that("plot_importance_ggpairs works for randomForest regression forests ", {
  p <- plot_importance_ggpairs(rf_r)
  expect_silent(suppressWarnings(suppressMessages(print(p))))
})

test_that("plot_importance_rankings works for randomForest regression forests ", {
  p <- plot_importance_rankings(rf_r)
  expect_silent(suppressWarnings(suppressMessages(print(p))))
})

# Test randomForest unsupervised forests ----------------------------------

test_that("measure_importance works for randomForest unsupervised forests", {
  imp_df <- measure_importance(rf_u, mean_sample = "all_trees",
                               measures = c("mean_min_depth", "accuracy_decrease", "gini_decrease",
                                            "no_of_nodes", "times_a_root", "p_value"))
  expect_equal(as.character(imp_df$variable),
               c("Petal.Length", "Petal.Width", "Sepal.Length", "Sepal.Width", "Species"))
})

test_that("important_variables works for randomForest unsupervised forests", {
  imp_vars <- important_variables(rf_u, k = 3,
                                  measures = c("mean_min_depth", "accuracy_decrease", "gini_decrease",
                                               "no_of_nodes", "times_a_root", "p_value"))
  expect_equal(imp_vars, c("Petal.Length", "Sepal.Length", "Species"))
})

test_that("min_depth_distribution works for randomForest unsupervised forests", {
  min_depth_dist <- min_depth_distribution(rf_u)
  expect_equal(
    min_depth_dist[min_depth_dist$tree == 1 & min_depth_dist$variable == "Sepal.Width", ]$minimal_depth,
    0)
})

test_that("min_depth_interactions works for randomForest unsupervised forests", {
  min_depth_int <- min_depth_interactions(rf_u, vars = c("Petal.Width"))
  # returns matrix
  expect_equal(
    min_depth_int[min_depth_int$interaction == "Petal.Width:Sepal.Length", ]$mean_min_depth,
    1,
    ignore_attr = "dim"
  )
})

test_that("plot_multi_way_importance works for randomForest unsupervised forests", {
  p <- plot_multi_way_importance(rf_u)
  expect_silent(suppressWarnings(print(p)))
})

test_that("plot_min_depth_distribution works for randomForest unsupervised forests", {
  p <- plot_min_depth_distribution(rf_u)
  expect_silent(print(p))
})

test_that("plot_min_depth_interactions works for randomForest unsupervised forests", {
  p <- plot_min_depth_interactions(rf_u)
  expect_silent(print(p))
})

test_that("plot_predict_interaction does not work for unsupervised forests", {
  expect_warning(p <- plot_predict_interaction(rf_u, data = iris, variable1 = "Petal.Width", variable2 = "Sepal.Length"))
  expect_null(p)
})

test_that("plot_importance_ggpairs works for randomForest unsupervised forests", {
  p <- plot_importance_ggpairs(rf_u)
  expect_silent(suppressWarnings(suppressMessages(print(p))))
})

test_that("plot_importance_rankings works for randomForest unsupervised forests", {
  p <- plot_importance_rankings(rf_u)
  expect_silent(suppressWarnings(suppressMessages(print(p))))
})