Just a note - it's usually most useful to look only at the 2-3 most recent CATs, especially if older ones are older than 4-6 weeks. When you go back too far, the data no longer actually reflects your current strengths and weaknesses.
For any of these weak areas, figure out:
(a) WHY you got something wrong, and / or
(b) WHY you took too long
If you don't know why, then you don't know what you need to do to try to improve. If you realize that you didn't know some rule or formula, you study it. If you realize that you don't feel comfortable doing certain math manipulations, you go drill / practice those things.
If you realize that you didn't really get what they were asking, then you seek out explanations that will help you learn how to better "decode" these questions and figure out what to do (and things like Thursdays with Ron can help with this, as well as articles on our blog - pay attention to when we talk about HOW TO KNOW that the problem is talking about a certain thing how to know to do a certain thing).
If you realize you made a careless mistake, then you figure out why you made the mistake and you implement some new good habit (or break some old bad habit) that will help you minimize this kind of careless mistake in future. Etc.
Exponent, roots, and percents are common, so spend more time there. Sets and probability are not very common, so get those wrong faster and spend the time elsewhere (I mean this in two senses: both during the test when you do get one and outside of the test when you're studying). Geo is of medium importance.
To address your "where to start?" issue, get some flash cards. On one side, write "when I see..." and on the other "I'll think / do..."
The first side will contain specific clues (if a problem contains several numbers / values such as 2, 3, 5, and 7... I'll notice that they all "happen" to be prime, so I'll suspect that the problem is likely about prime!) or actual text from a question but NOT the entire question - just the relevant text that could appear in many questions (eg, when I see >0 or <0 in a quant problem, I'll suspect that the problem is really about positives and negatives and try to approach it from that direction).
Awesome on SC.
CR weaken:
http://www.manhattangmat.com/blog/index ... n-problem/RC infer:
http://www.manhattangmat.com/articles/a ... estion.cfm