Decostand "frequency"

[invertdna]

decostand provides the method "frequency" as an option for standardization, and I think the helpfile might be incorrect:

"frequency: divide by margin maximum and multiply by the number of non-zero items, so that the average of non-zero entries is one (Oksanen 1983; default MARGIN = 2)."

But carrying out the operations as described does not yield a mean of one. For example:

a <- sample(0:9, 100, replace = T)  #create some random data; note that it doesn't matter whether these are integers or decimals, etc. 
b <- a/max(a)     #divide by margin max
d <- b*sum(b!=0)   #multiply by number of non-zero items
mean(d[d!=0])   #take mean of non-zero items

yields an integer between about 45 and 55, depending on the sampling.

By contrast,

f <- decostand(a, method = "frequency") 
mean(f[f!=0])

does have the intended effect of making the mean of non-zero numbers equal to zero.

The code for decostand has the following:

frequency = {
        if (missing(MARGIN)) MARGIN <- 2
        tmp <- pmax(k, apply(x, MARGIN, sum, na.rm = na.rm))
        fre <- apply(x > 0, MARGIN, sum, na.rm = na.rm)
        tmp <- fre/tmp
        x <- sweep(x, MARGIN, tmp, "*")
    }

Which essentially divides the number of non-zero entries by the vector SUM, and then multiplies the vector by that dividend.

In sum, I think the helpfile wrongly describes the operation performed.

[jarioksa]

Corrected in commit 6617fca. Thanks.