Tabulate missing values

Summarize missing values by variable (misstable summarize) or tabulate missingness patterns across rows (misstable patterns), following Stata's misstable command.

Usage

misstable(.data, ..., .by = NULL, .patterns = FALSE, .all = FALSE)

Arguments

.data

A data frame, tibble, or data.table.

...

<tidy-select> or <data-masking> Columns to select or compute with select_or_transmute(). Existing columns, selection helpers, and named expressions can be mixed. Omit to check all columns except grouping keys.

.by

<tidy-select> Optional grouping variables. Grouped data use their groups unless .by is supplied. Select existing columns only; create new grouping columns with dplyr::mutate() first.

.patterns

If TRUE, tabulate missingness patterns instead of summarizing individual variables.

.all

If TRUE, include variables without missing values in either mode. Defaults to FALSE, as in Stata.

Value

An ungrouped tibble with any grouping keys and:

• By default, one row per checked variable and group, with variable, miss_n (missing count), miss_pct (missing percentage), and valid_n (observed count). No checked variables gives zero rows.

• With .patterns = TRUE, one row per observed pattern and group, with a 1 (observed) or 0 (missing) column per checked variable, plus n (frequency) and percent. With no checked variables, each nonempty group has one row containing only its keys, n, and percent.

Percentages range from zero to 100 within each group. Summary rows for zero-row ungrouped inputs have zero counts and NA percentages when .all = TRUE; empty inputs have no pattern rows.

Details

Atomic missingness follows is.na(), including NaN, haven's tagged missing values, and SPSS user-missing codes and ranges. In list columns, NA is missing but NULL is observed. Matrix and data-frame columns are missing only when all components in a row are missing, following vctrs conventions.

Computed columns are evaluated on ungrouped working data before missingness is summarized within groups. Create columns with grouped dplyr::mutate() first when a calculation needs to run separately within each group.

Empty strings are treated as observed values, unlike in Stata. Use na_if_value() to convert them to missing values.

With .all = FALSE, variables are checked only if they contain missing values anywhere in the input. A message is displayed when none do. Summary rows follow selection order within each group; patterns follow decreasing frequency, with ties preferring observed values column by column. Grouping keys must not conflict with output column names. Pattern variables must also differ from grouping keys and the names n and percent.

See also

Other inspection functions: browse(), codebook(), names, var_type(), variables()

Examples

# Summarize missing values
misstable(lifeexp)
#> # A tibble: 2 x 4
#>   variable  miss_n miss_pct valid_n
#>   <chr>      <int>    <dbl>   <int>
#> 1 gnppc          1     1.47      67
#> 2 safewater     14    20.6       54

misstable(lifeexp, .all = TRUE)
#> # A tibble: 6 x 4
#>   variable  miss_n miss_pct valid_n
#>   <chr>      <int>    <dbl>   <int>
#> 1 region         0     0         68
#> 2 country        0     0         68
#> 3 popgrowth      0     0         68
#> 4 lexp           0     0         68
#> 5 gnppc          1     1.47      67
#> 6 safewater     14    20.6       54

# Tabulate missingness patterns
misstable(lifeexp, .patterns = TRUE)
#> # A tibble: 3 x 4
#>   gnppc safewater     n percent
#>   <int>     <int> <int>   <dbl>
#> 1     1         1    53   77.9 
#> 2     1         0    14   20.6 
#> 3     0         1     1    1.47

# Select variables and group observations
misstable(lifeexp, gnppc, safewater, .by = region)
#> # A tibble: 6 x 5
#>   region               variable  miss_n miss_pct valid_n
#>   <dbl+lbl>            <chr>      <int>    <dbl>   <int>
#> 1 1 [Europe & C. Asia] gnppc          1     2.27      43
#> 2 1 [Europe & C. Asia] safewater      2     4.55      42
#> 3 2 [North America]    gnppc          0     0         14
#> 4 2 [North America]    safewater      8    57.1        6
#> 5 3 [South America]    gnppc          0     0         10
#> 6 3 [South America]    safewater      4    40          6

# Check an existing column alongside a computed column
misstable(lifeexp, gnppc, water_share = safewater / 100, .all = TRUE)
#> # A tibble: 2 x 4
#>   variable    miss_n miss_pct valid_n
#>   <chr>        <int>    <dbl>   <int>
#> 1 gnppc            1     1.47      67
#> 2 water_share     14    20.6       54