Compute for a data sample.
Employ Rousseeuw and Croux's finite sample size correction.
Let's replace four of the values with very large values.
The value of stays bounded, because it has a high breakdown point.
Compute for a normal distribution.
The symbolic result is an expression involving the inverse (see RootOf) of the error function (see erf). It evaluates to the same floating-point number.
Generate a random sample of size 1000000 from the same distribution and compute the sample's .
Consider the following Matrix data set.
We compute for each of the columns.