When it comes to table selection in cash games the most important aspect is to find people, who play to many hands, and ideally then get position on them. The last part is very important, because if they have position on you, you are going to be playing a lot of pots out of position, which is a big disadvantage, even if you have a stronger hand on average.
So basically the most important stat for table selection is VPIP. A full ring cash table should ideally have players with a VPIP above 40% or at the bare minimum several players with a VPIP between 25% and 40%, and for 6-max you want to raise those numbers a bit.
Personally I label players with a VPIP above 40 yellow, and you could call this "fish". Anywhere between 20-40% (for full ring) can either be a LAG, which I label blue or a semi loose passive (SLP), which I label orange. This depend on the gap between VPIP and PFR. A 25/21 (full ring) is a LAG, but a 25/9 is an SLP. SLPs are usually bad regs. Its people, who are transforming from a fish to a solid and possibly winning player, but they are only sort of half way there.
Now its important to actually understand, what these labels or player types mean, and what kind of adjustments you want to make. Its a pretty common mistake for people to get completely out of line, because "LOL he is just a fish", but you need to think about, what the numbers actually mean. If for instance you get 3-bet by someone with a VPIP / PFR of 52/7, then you need to take it serious, because chances are, he has a very good hand. If he had some random junk, he would just have called you, like he did all those other times.
Finally some people will make a distinction between "agro fish" and "passive fish", and the first category is also called "maniacs". I personally dont do that with my labels, but I certainly dont respect a raise as much, if its coming from a 53/36, as if it comes from a 53/6.
The point being that you need to look at, what people are actually doing, and not just "LOL its a fish". This also includes looking at how often or rarely they fold to C-bets, when the sample gets big enough, how often they donk bet and so on and so forth. Some fish might donk bet 10% and others 65%, and again this means something different, even though their VPIP is identical.