Counts how often each value of a column occurs, with percentages of all
rows and of the non-missing rows, and missing values counted as their
own row. It is the first look at a categorical variable (a charge type,
a region, a response category), and the counterpart of
janitor::tabyl().
Arguments
- data
A data frame, or a vector.
- col
Column name (a string), when
datais a data frame; ignored for a vector.- maxlvl
Maximum value rows to return, most frequent first (default 25). The remainder are folded into one
"(other)"row so the percentages still sum to 100.- sort
Logical: sort by descending count (default
TRUE);FALSEkeeps the sorted order of the values.
Value
A data frame of class c("bricklayer_freq", "data.frame")
with columns value (character), n (integer count),
pct (numeric, percent of all rows) and pctvalid (numeric,
percent of non-missing rows; NA on the missing row).
Details
Two percentage columns, because both questions get asked and conflating
them is how missingness gets hidden: pct is the share of ALL
rows, pctvalid the share of rows where the value is present.
When a column is 40% missing those two differ enormously, and only the
second describes the values that are actually there.
Values are tabulated as text (as.character()), so numbers and
factors are counted by their printed value, and factor levels that do not
occur get no row. Rows are sorted by descending count (ties in the order
table() gives) or kept in sorted value order with
sort = FALSE. Beyond maxlvl distinct values, the rest are
folded into one "(other)" row so the percentages still sum to 100.
The missing row, when there is one, comes last, with value NA
and pctvalid NA. An empty input gives a table with no rows.
The print method adds a bar column.
Errors: a data frame without col; a col that is not
in data; maxlvl not one positive whole number.
See also
profcols() for every
column at once.
Examples
df <- data.frame(grade = c("a", "b", "b", "c", NA, "b"),
stringsAsFactors = FALSE)
freqtbl(df, "grade")
#> ╭── Frequency table ──────────────────────────────────────────────────────╮
#> │ Values 6 │
#> │ Levels shown 3 │
#> │ Missing 1 │
#> │ │
#> │ value n pct pctvalid bar │
#> │ ----------- - ----- -------- -------------------- │
#> │ b 3 50.00 60 #################### │
#> │ a 1 16.70 20 ####### │
#> │ c 1 16.70 20 ####### │
#> │ – (missing) 1 16.70 ####### │
#> │ │
#> │ pct is the share of all values, pctvalid the share of the non-missing │
#> │ ones; the bar is scaled to the commonest level. │
#> ╰─────────────────────────────────────────────────────────────────────────╯
# The two percentage columns differ exactly by the missingness.
freqtbl(df, "grade")[, c("pct", "pctvalid")]
#> ╭── Frequency table ────────────────────────────────────────────────────╮
#> │ Rows 4 │
#> │ Columns 2 │
#> │ │
#> │ pct pctvalid │
#> │ ----- -------- │
#> │ 50.00 60 │
#> │ 16.67 20 │
#> │ 16.67 20 │
#> │ 16.67 │
#> │ │
#> │ Some of this result's columns are not here (a subset of it?), so it │
#> │ is shown as a plain table. │
#> ╰───────────────────────────────────────────────────────────────────────╯
# A vector works directly.
freqtbl(c(1, 1, 2, 3, 3, 3))
#> ╭── Frequency table ──────────────────────────────────────────────────────╮
#> │ Values 6 │
#> │ Levels shown 3 │
#> │ Missing 0 │
#> │ │
#> │ value n pct pctvalid bar │
#> │ ----- - ----- -------- -------------------- │
#> │ 3 3 50.00 50.00 #################### │
#> │ 1 2 33.30 33.30 ############# │
#> │ 2 1 16.70 16.70 ####### │
#> │ │
#> │ pct is the share of all values, pctvalid the share of the non-missing │
#> │ ones; the bar is scaled to the commonest level. │
#> ╰─────────────────────────────────────────────────────────────────────────╯
# Sorted value order rather than frequency order.
freqtbl(c("c", "a", "b", "a"), sort = FALSE)
#> ╭── Frequency table ──────────────────────────────────────────────────────╮
#> │ Values 4 │
#> │ Levels shown 3 │
#> │ Missing 0 │
#> │ │
#> │ value n pct pctvalid bar │
#> │ ----- - --- -------- -------------------- │
#> │ a 2 50 50 #################### │
#> │ b 1 25 25 ########## │
#> │ c 1 25 25 ########## │
#> │ │
#> │ pct is the share of all values, pctvalid the share of the non-missing │
#> │ ones; the bar is scaled to the commonest level. │
#> ╰─────────────────────────────────────────────────────────────────────────╯
# Long tails are folded so the percentages still total 100.
set.seed(1)
ft <- freqtbl(sample(letters, 500, TRUE), maxlvl = 5)
ft
#> ╭── Frequency table ──────────────────────────────────────────────────────╮
#> │ Values 500 │
#> │ Levels shown 6 │
#> │ Missing 0 │
#> │ │
#> │ value n pct pctvalid bar │
#> │ ------- --- ------ -------- -------------------- │
#> │ y 25 5.000 5.000 # │
#> │ e 24 4.800 4.800 # │
#> │ t 24 4.800 4.800 # │
#> │ a 23 4.600 4.600 # │
#> │ f 23 4.600 4.600 # │
#> │ (other) 381 76.200 76.200 #################### │
#> │ │
#> │ pct is the share of all values, pctvalid the share of the non-missing │
#> │ ones; the bar is scaled to the commonest level. │
#> ╰─────────────────────────────────────────────────────────────────────────╯
sum(ft$pct)
#> [1] 100
# a data frame needs a column name
try(freqtbl(df))
#> Error : `col` is required when `data` is a data frame