Smart row-wise sum with missing data handling and type preservation

Computes the row-wise sum across multiple input vectors, while allowing control over how many non-missing values must be present for a valid result. If the number of non-NA values in a row is below min_present, the result is set to NA of the appropriate type.

Usage

fallback_row_sum(..., min_present = 1, .keep_zero_as_zero = TRUE)

Arguments

...: Numeric or integer vectors of equal length to be summed row-wise.
min_present: Integer. Minimum number of non-NA values required per row to return a sum. Rows with fewer than min_present non-NA values return NA_integer_ if all inputs are integer, otherwise NA_real_.
.keep_zero_as_zero: Logical. Currently unused, reserved for future development. Defaults to TRUE.

Value

An integer or numeric vector of row-wise sums, depending on the input types, with appropriate NA values where insufficient data are present.

Details

The return type matches the inputs:

If all inputs are integer vectors, the result is an integer vector with NA_integer_ where insufficient data are present.
Otherwise, the result is a numeric (double) vector with NA_real_ where insufficient data are present.

Examples

# all integer inputs -> integer output
x <- c(1L, 2L, NA_integer_)
y <- c(3L, NA_integer_, 4L)
fallback_row_sum(x, y)
#> [1] 4 2 4
typeof(fallback_row_sum(x, y))  # "integer"
#> [1] "integer"

# mixed integer and numeric inputs -> numeric output
z <- c(1, 2, NA)
fallback_row_sum(x, z)
#> [1]  2  4 NA
typeof(fallback_row_sum(x, z))  # "double"
#> [1] "double"

# using min_present to control NA behaviour
fallback_row_sum(c(1, NA), c(2, NA), min_present = 1)
#> [1]  3 NA