In this area (versus something more formal like determining ‘circular error probable’), I look at ~80% of the shots. Maybe ‘most’ is better than some number. Fliers are gonna fly; don’t stress about them. If you were going to chunk that shot, you were going to chunk it regardless of your aim point or course management, so I don’t let those instances go into my determination of what my actual future dispersion will be. IOW, I chuck the outliers from my data set. Looks like you do too in your example.
For your 6 iron, if you were to plot each of the hits, from the 90 percent of the shots you’re evaluating, would there be clusters to the misses? For most people, that shot map won’t be circular and random. It’ll be biased to one side, and it will be extremely biased to closer vs further. Which is where Fawcett’s take enough club for the back of the green, comes in.
You at least, have a sim that can show you graphically, where all of those shots went. And can, in theory, take an overlay of a green that you play, and move that shot distribution around until you fit as many as you’re going to on the green. (With exceptions of course, if that particular aim point that maximizes total balls on green, leads to an unacceptable number of misses in a penalty area. Or taking a slightly lower number of balls on the green, if that leads to more balls that go off the green being in much easier recovery positions. And so on.)
It may be that you’re only going to fit, say 45% of the shots on the green, no matter the aim point. Which is fine; you’re still getting the most out of what you have. I’m sure you’ve already tried this, but for others, maybe a “80-90” percent iron swing gets you much better distance and spread control, even if you consequently lose 10-15 of distance? So what if I now have to take a 4H, instead of a 6I, into this green, if my dispersion tightens up from 25 yds of distance to 10, and 60 yds L-R to 40?
Or maybe that doesn’t work for you at all. But you can find out, with the monitor equipment you have. Just make sure you’re also getting data for grass lies too.