Hey kerrymaid.

Are you talking about this is a probabilistic/expected-frequency kind of way?

Basically if you have a ratio of tx/rx then it means that you will need to send round(tx/rx) packets on average to get a received packet. Note that I say "on average" because the probability represents an infinite limit of the ratio between tx packets and rx packets. If you have additional information about the distribution of rx to tx (or vice versa) then you can use that PDF to get the probability for some interval. (Also when I say round I mean that you need to add 1 if you have a fractional answer to account for integer values that make sense).

You can use the fact that the distribution of the mean of any distribution will be approximately normal: use this if you want to construct a confidence interval as opposed to looking at a point estimate (which is what your table of values contain).