This question was inspired by a Stack Overflow question forwarded to me by Matt Simoneau.

Given a vector x, make an indexed probability vector p that shows the likelihood of each unique element of x occurring. You can assume that x will always be composed of integers greater than 0.

For example, if x is [4 4 4 4] then the number 4 is 100% likely to occur and no other number has any likelihood of occurring. By convention, p has as many elements as the highest number in x. This makes the indexing straightforward. So in this case, p = [0 0 0 1]. That is, p(4) = 1, which is to say that 4 occurs with 100% probability, as noted above. Note that p has no dependence on the order of the elements in x.

If

 x = [1 2 1 2 1 2 1 2 1 2] 

then

 p = [0.5 0.5].

And if

 x = [5 5 2 8] 

then

 p = [0 0.25 0 0 0.5 0 0 0.25]