First, Trackman does not capture carry distance, total distance, etc. directly. It estimates those variables from the what it does measure directly-- clubhead speed, ball speed, angle of attack, face to path, spin rate, spin axis, etc.
Second, "best", "longest", and "least dispersion" are not likely to all be the same thing. For a driver, sure it makes plenty of sense to look at longest and for many people that may mean best. But I think to approach anything resembling "best" you would need to define a proxy variable similar to strokes gained. MyGolfSpy already does this with the majority of their testing. I would check out how they have defined those things. Even least dispersion is not necessarily as simple as it sounds. If a club has a tendency to miss left then it's a non-starter in my bag, even if it's overall low dispersion. My ability to navigate a difficult course depends on keeping my misses in check which means managing where they go (i.e. right of the target, not left of the target). So I don't care if a club is low dispersion if it misses left-- it's not going in my bag.
Third, are the golfers in this test dataset properly fitted with the clubs? For example, the "best" head for an individual golfer depends on a proper shaft fit and vice versa. If they are not, you would need to account for this.
Lastly, you need to think about confounding variables such as swing speed (and/or handicap) for this analysis for the conclusions to make any sense. The "best" ball for example is much different for me than a 75 year old woman who needs lots of distance help. For me, the best ball is likely some prov1 variant, chrome soft, TP5, etc. In general, I have plenty of distance so I need something which minimizes dispersion and has good greenside spin control. For our hypothetical 75 year old woman it's going to be the ball that goes the furthest, since she doesn't generate enough clubhead speed for greenside spin control anyway.
Source; PhD statistician and <5 handicap golfer.