inst/tinytest/test-datasummary_skim.R

In modelsummary: Summary Tables and Plots for Statistical Models and Data: Beautiful, Customizable, and Publication-Ready

dat <- mtcars
dat$vs <- as.logical(dat$vs)
dat$gear <- as.factor(dat$gear)

# basic
expect_silent(datasummary_skim(dat, type = "numeric", output = "data.frame"))
tmp <- suppressWarnings(datasummary_skim(dat, type = "numeric", output = "data.frame"))
expect_equivalent(dim(tmp), c(9, 9))

tmp <- datasummary_skim(dat, type = "categorical", output = "data.frame")
expect_equivalent(dim(tmp), c(5, 4))

# display NA in categorical
tmp <- mtcars
tmp$vs <- as.logical(tmp$vs)
tmp$vs[1:3] <- NA

tab <- datasummary_skim(tmp, type = "categorical", output = "data.frame")
expect_equivalent(nrow(tab), 3)

tmp$vs <- factor(tmp$vs)
tab <- datasummary_skim(tmp, type = "categorical", output = "data.frame")
expect_equivalent(nrow(tab), 2)

# simple categorical
tab <- datasummary_skim(dat, type = "categorical", output = "data.frame")
expect_inherits(tab, "data.frame")
expect_equivalent(dim(tab), c(5, 4))

# no numeric variables
tab <- datasummary_skim(tmp)
expect_inherits(tab, "tinytable")

tmp <- data.frame(
# all fully missing
  a = rep(NA_real_, 20),
  b = rep(NA_character_, 20))
expect_error(datasummary_skim(tmp),
  pattern = "all numeric")

# too many columns
tmp <- data.frame(matrix(rnorm(252 * 3), ncol = 252))
expect_error(datasummary_skim(tmp),
  pattern = "more than 250 variables")


# errors and warnings: categorical

# no categorical
tmp <- data.frame(a = rnorm(10))
expect_error(datasummary_skim(tmp, type = "categorical"),
  pattern = "contains no logical")

# too many columns
tmp <- data.frame(matrix(as.character(rnorm(52 * 3)), ncol = 52))
expect_error(datasummary_skim(tmp, type = "categorical"),
  pattern = "more than 50 variables")

# fmt
known <- c("33.9000", "8.0000", "472.0000", "335.0000", "4.9300", "5.4240", "22.9000", "1.0000", "8.0000")
tmp <- datasummary_skim(dat, output = "data.frame", type = "numeric", fmt = "%.4f")
expect_equivalent(tmp$Max, known)

# empty string factor level (tricky for tables::tabular)
tmp <- data.frame(a = c(rep("", 3), rep("b", 5), rep("c", 10)))
tmp <- datasummary_skim(tmp, type = "categorical", output = "data.frame")
expect_equivalent(dim(tmp), c(3, 3))

# too many factor levels
N <- 20000
tmp <- data.frame(
  a = sample(letters, N, replace = TRUE),
  b = as.character(sample(1:100, N, replace = TRUE)))
expect_warning(datasummary_skim(tmp, type = "categorical", output = "data.frame"))
tmp <- suppressWarnings(datasummary_skim(tmp, type = "categorical", output = "data.frame"))
expect_equivalent(ncol(tmp), 3)

# completely missing variables are dropped
tmp <- dat
tmp$junk <- rep(NA_character_, nrow(dat))
tmp <- expect_warning(datasummary_skim(tmp, type = "categorical", output = "data.frame", histogram = FALSE))
expect_false("junk" %in% tmp[[1]])

# simple dataset
tab <- datasummary_skim(mtcars, type = "dataset", output = "data.frame")
expect_equivalent(dim(tab), c(4, 2))

# tibble input does not error
penguins <- read.csv("https://vincentarelbundock.github.io/Rdatasets/csv/palmerpenguins/penguins.csv")
penguins <- tibble::as_tibble(penguins)
tab <- datasummary_skim(penguins)
expect_inherits(tab, "tinytable")


# no validity; just no error
tmp <- mtcars |>
  transform(cyl = factor(cyl), gear = factor(gear)) |>
  subset(select = c("cyl", "mpg", "hp", "vs", "gear"))
x <- datasummary_skim(tmp, type = "all", output = "latex")
y <- datasummary_skim(tmp, type = "numeric")
z <- datasummary_skim(tmp, type = "categorical")
expect_inherits(x, "tinytable")
expect_inherits(y, "tinytable")
expect_inherits(z, "tinytable")


# by is non-numeric
tmp <- mtcars |>
  transform(cyl = factor(cyl), gear = factor(gear)) |>
  subset(select = c("cyl", "mpg", "hp", "vs", "gear"))
tab <- datasummary_skim(tmp, by = "gear", type = "all")
expect_inherits(tab, "tinytable")


# Issue #627: histograms in gt
# expect_inherits(datasummary_skim(mtcars, output = "gt"), "gt_tbl")


# # Big refactor
# Q
# pkgload::load_all()
# fun <- list(
#             "mean" = mean,
#             "sd" = sd,
#             "Histogram" = function(x) ""
# )
# mtcars[, c("qsec", "mpg", "drat", "vs")] |>
#   datasummary_skim(fmt = 1, by = "vs", fun_numeric = fun, output = 'html') #|>
#   # print("html")
#   # print("html")
#   #
#


# # RDatasets tests: must be commented out
# any_categorical <- function(x) {
#   # datasummary_skim ignores characters with more than 10 levels
#   for (n in colnames(x)) {
#     if (is.character(x[[n]]) | is.logical(x[[n]])) {
#       bad1 <- length(unique(x[[n]])) > 20
#       bad2 <- all(is.na(x[[n]]))
#       flag <- bad1 | bad2
#       if (flag) {
#         x[[n]] <- NULL
#       }
#     }
#   }
#   idx <- sapply(x, function(x) 
#                 is.factor(x) | is.logical(x) | is.character(x))
#   any(idx) & (ncol(x) < 40)
# }
# any_numeric <- function(x) {
#   any(sapply(x, function(x) is.numeric(x))) & (ncol(x) < 51)
# }
# csv_files <- Sys.glob("~/Downloads/Rdatasets/csv/*/*.csv")
# for (f in csv_files) {
#   dat <- read.csv(f)
#   if (any_numeric(dat)) {
#     test_that(paste("numeric:", f), {
#       expect_error(datasummary_skim(dat, type="numeric"), NA)
#     
#   }
#   if (any_categorical(dat)) {
#     test_that(paste("categorical:", f), {
#       expect_error(datasummary_skim(dat, type="categorical"), NA)
#     
#   }
# }