I have a set of X observations that reflect the number of times each combination of 2 values have been seen. E.g. if I pick a shape and a color, I have seen certain shape+color combinations but others - which I know may exist - I have never seen before.
I am trying to answer the question: if I pick another combination of shape+color what is the probability that I will have seen it before?
My confusion has to do with assigning probabilities to observations that I have never seen before: for example, if I have never seen a green square, it doesn't mean that the probability is zero ...
How do I use the number of observations I made (X) and the number of possible combinations (Y) to compute the probability for events that I have never seen before but which nonetheless as possible?
