What makes a post
overrated and how can I "fix" that?
Much of that page has secret tooltips to explain what fields mean. Underrated is "no votes in the last year, helpful", and overrated is "no votes in the last year, unhelpful". It's so perfectly clear now there's no way anyone could need further explanation!
...but just in case: underrated posts are posts that have been marked helpful, but haven't gotten any upvotes in the last year, even though they clearly deserve them since they've been helping people. Overrated posts have been getting marked unhelpful, but haven't gotten any downvotes in the last year.
As for "fixing" this, it often requires cleaning up answers that have been marked helpful and then upvoting them. Theoretically, moderators and top users are working to improve and upvote helpful posts, and downvote unhelpful posts.
post feedback breakdown actually indicate/measure?
This has a tooltip too: "How many feedback votes did posts get?". The top row is the amount of feedback a post has, and the bottom row is how many posts have that much feedback, so in your screenshot 595437 posts on SO (77.3% of the total) have exactly 1 feedback vote. 4581 posts (0.6%) have 10 or more
How should I interpret the differences between anonymous and registered up/down feedback?
The differences between anonymous and registered feedback is just to see if drive-by users are finding posts helpful, since they can't vote otherwise. And I guess to see how many anonymous users are taking advantage of the feature, since I think it was developed specifically for them
What does the
post score to feedback correlation chart thing even mean?
The top row is the post's net score (that is, net of up and down votes), and the bottom row is how many posts with that score have gotten feedback (I think). So 16612 posts with a negative score have gotten feedback. Correlating post score and feedback lets you see how good the site is at voting; posts that get the most feedback are probably the most useful (they're at least stumbled upon the most), so they should have the most votes. It seems to count both positive and negative feedback equally though, which diminishes the usefulness a bit -- posts with highly negative feedback shouldn't have high scores