These two functions can be used to calculate the mode (most frequently observed value) of a sample, and the actual frequency of the modal value. The only complication is in respect to missing data. If na.rm = FALSE
, then there are multiple possibilities for how to calculate the mode. One possibility is to treat NA
as another possible value for the elements of x
, and therefore if NA
is more frequent than any other value, then NA
is the mode; and the modal frequency is equal to the number of missing values. This is the version that is currently implemented.
Another possibility is to treat NA
as meaning "true value unknown", and to the mode of x
is itself known only if the number of missing values is small enough that -- regardless of what value they have -- they cannot alter the sample mode. For instance, if x
were c(1,1,1,1,2,2,NA)
, we know that the mode of x
is 1
regardless of what the true value is for the one missing datum; and we know that the modal frequency is between 4 and 5. This is also a valid interpretation, depending on what precisely it is the user wants, but is not currently implemented.
Because of the ambiguity of how na.rm = FALSE
should be interpreted, the default value has been set to na.rm = TRUE
, which differs from the default value that I've used elsewhere in the package.