revdep/library.noindex/vip/old/vip/tinytest/test_pkg_C50.R

In AFIT-R/vip: Variable Importance Plots

# Exits
if (!requireNamespace("C50", quietly = TRUE)) {
  exit_file("Package C50 missing")
}

# # Load required packages
# suppressMessages({
#   library(C50)
# })

# Generate Friedman benchmark data
friedman2 <- gen_friedman(seed = 101, n_bins = 2)

# Fit model(s)
fit1 <- C50::C5.0(y ~ ., friedman2)
fit2 <- C50::C5.0(x = friedman2[, paste0("x", 1L:10L)], y = friedman2$y)

# Compute VI scores
vis1 <- vi_model(fit1)
vis2 <- vi_model(fit2)

# Expectations for `vi_model()`
expect_identical(
  current = vis1,
  target = vis2
)
expect_identical(
  current = vis1$Importance,
  C50::C5imp(fit1, metric = "usage")$Overall
)
expect_identical(
  current = vi_model(fit1, type = "splits", pct = FALSE)$Importance,
  C50::C5imp(fit1, metric = "splits", pct = FALSE)$Overall
)

# Expectations for `get_training_data()`
expect_identical(
  current = vip:::get_training_data.C5.0(fit1),
  target = friedman2
)
expect_identical(
  current = vip:::get_training_data.C5.0(fit2),
  target = subset(friedman2, select = -y)
)

# Expectations for `get_feature_names()`
expect_identical(
  current = vip:::get_feature_names.C5.0(fit1),
  target = paste0("x", 1L:10L)
)
expect_identical(
  current = vip:::get_feature_names.C5.0(fit2),
  target = paste0("x", 1L:10L)
)

# Call `vip::vip()` directly
p <- vip(fit2, method = "model", type = "splits", pct = FALSE,
         include_type = TRUE)

# Expect `p` to be a `"gg" "ggplot"` object
expect_identical(
  current = class(p),
  target = c("gg", "ggplot")
)

# Display VIPs side by side
grid.arrange(
  vip(vis1, include_type = TRUE),
  vip(vis2, include_type = TRUE),
  p,
  nrow = 1
)