Frecency is a portmanteau of "frequency" and "recentness."
I don't know the exact scoring mechanism but potential matches are scored by both frequency of access and recentness of access. Firefox uses this for matches in the URL bar, and I think other browsers do as well.
For `z`, as an example, suppose I have two paths:
~/projects/fun_stuff_1
~/projects/fun_stuff_2
`fun_stuff_1` has been accessed 100 times, but all of those accesses were over two years ago.
`fun_stuff_2` has been accessed only once, but it was five minutes ago.
So if I type `$ z fun_stuff` or even `$z fun` it's going to rank `fun_stuff_2` ahead of `fun_stuff_1` in the possibilities. In practice, while obviously not perfect, frecency tends to match the user's intent a surprisingly high amount of the time... a good frecency implementation is a delight for the user, honestly.
z works a bit like a web browser does when autocompleting URLs from history rather than a file system search tool like find. z “learns” the file system paths as you traverse them in a terminal.
Could you maybe give a few examples of how you might find something eg. using find(1), and then how you could find it using z?