Uh, that's what I've been doing, but that's not what is occurring. As I've posted, even adding a single new characteristic has the potential to radically re-shape the tree. Which occurred when I added drive type as a characteristic, which was the post you quoted.
As another example here are two more trees. The first based solely on engine and fuel characteristics. The second based on physical body design and features. The two trees are completely different.
When I calculate a P-value for these trees, the result is P of ~1. This means there is no correlation between the two trees.
![]()
![]()
It's also worth noting that your own attempts to sort vehicles by physical size don't bear out in any of the results I've calculated. While I have yet to include gross physical size as a measurement, the trees I've computed so far show little convergence based on size. Just the varied locations of vehicles like the small cars, vans, trucks, and SUVs confirm that.
What I'm finding is that vehicle characteristics appear to be highly independent of each other. Which makes sense given these are manufactured objects that don't have hereditary constraints.
ok. notice that the bicycle remain in the same position no matter what. its fit well with what i said about general similarity, since most cars shared most parts with other cars rather then with other groups like a bicycle. so we still stay with a single basic hierarchical tree- bicycles, cars, and trucks.
Upvote
0