Logical OR with NA tolerance

Compute a position-wise logical OR across multiple logical inputs. When no TRUE is present, it allows up to a specified fraction of missing values per position before returning NA.

Usage

or(..., .na_allowed = 0.2)

Arguments

...

Logical vectors of equal length, or a single logical matrix/data.frame. Matrix/data.frame inputs must have at least one column. Inputs are coerced to logical via as.logical, preserving NA.

.na_allowed

(Scalar numeric: 0.2; [0, 1])
Maximum allowed fraction of NA per position before the result is set to NA when no TRUE is present.

Value

A logical vector with values TRUE, FALSE, or NA per the rule in Details.

Details

Standard logical disjunction over multiple vectors (e.g., x | y | z). Each output position corresponds to one vector index, or to one row when a matrix or data.frame is supplied. Returns TRUE for a position if at least one element is TRUE, regardless of missing values. Returns NA for a position if no TRUE is present and the fraction of NA exceeds .na_allowed. Otherwise, missing values are treated as FALSE when evaluating the OR.

For each position (row) \(i\) across \(p\) inputs, let \(m_i\) be the number of NAs and \(t_i\) be the number of TRUEs. Define the missing fraction \(r_i = m_i / p\). Then: $$ \text{or\_na}_i = \begin{cases} \text{TRUE}, & \text{if } t_i \ge 1, \\ \mathrm{NA}, & \text{if } t_i = 0 \text{ and } r_i > \text{.na\_allowed}, \\ \text{FALSE}, & \text{if } t_i = 0 \text{ and } r_i \le \text{.na\_allowed}. \end{cases} $$

Notes:

1. When .na_allowed = 0, behavior matches strict OR: NA propagates only when no TRUE is present.

2. When \(\text{.na\_allowed} = 1\), missingness never forces NA. NAs are treated as FALSE, so the result is TRUE iff any TRUE is present, otherwise FALSE.

See also

base::Logic

Examples

#----------------------------------------------------------------------------
# or() examples
#----------------------------------------------------------------------------
library(bkbase)

x <- c(FALSE, FALSE, FALSE, TRUE,  FALSE)
y <- c(FALSE, FALSE, NA,    FALSE, NA)
z <- c(FALSE, NA,    NA,    NA,    TRUE)

# Base R strict OR
x | y | z
#> [1] FALSE    NA    NA  TRUE  TRUE

# NA tolerance at 20%
or(x, y, z, .na_allowed = 0.2)
#> [1] FALSE    NA    NA  TRUE  TRUE

# NA tolerance at 40%
or(x, y, z, .na_allowed = 0.4)
#> [1] FALSE FALSE    NA  TRUE  TRUE

# Single data.frame input
df <- data.frame(x = x, y = y, z = z)
or(df, .na_allowed = 0.4)
#> [1] FALSE FALSE    NA  TRUE  TRUE

# TRUE dominates even when NA is present
or(TRUE, NA, .na_allowed = 0)
#> [1] TRUE