These data are just for actual wait times where we have valid, posted wait times within 15 minutes on both sides, from the Animal Kingdom. So yeah, a subset of the total number of actual wait times we have.
There is a lot of noise in the posted data, for various definitions of 'noise'. We've talked about it in previous blog posts. Some of the noise demonstrates group behavior that's physically impossible in real-world conditions.
Yes, they should be able to do much, much better. But they're not doing much, much better, so the question is why? My bet is still that nobody high enough in the organization has been made aware of it yet.