NA-aware Row-wise Functions

Drop-in replacements for rowMeans() and rowSums() that default to na.rm = TRUE. Both return NA for rows where ALL values are missing (base rowMeans() returns NaN, base rowSums() returns 0).

Usage

rowMeans(x, na.rm = TRUE, all_na = NULL, dims = 1L, ...)

rowSums(x, na.rm = TRUE, all_na = NULL, dims = 1L, ...)

Arguments

x

A numeric matrix or data frame.

na.rm

Logical. Should missing values be removed? Default TRUE.

all_na

Character. What to do when all values are NA: "error" (default) throws an error, "base" returns what base R does with na.rm = TRUE (NaN for rowMeans(), 0 for rowSums()), "na" returns NA. If NULL, uses getOption("tidyna.all_na", "error").

dims

Integer. Number of dimensions to treat as rows.

...

Additional arguments passed to the base function.

Value

A numeric or complex array of suitable size, or a vector if the result is one-dimensional.

Examples

mat <- matrix(c(1, NA, 3, NA, NA, NA), nrow = 2, byrow = TRUE)
rowSums(mat)
#> Warning: ⚠️ 4 missing values removed.
#> Warning: ⚠️ 1 row had all NA values.
#> [1]  4 NA

# Compare to base R:
base::rowSums(mat, na.rm = TRUE)
#> [1] 4 0