I would strongly agree about the

error log.

I first started to have 4/20 correct ( almost nothing) and know I have 15/20 correct.

I noticed that a limited number of errors are tested again and again. ( like the only prime number is 2; that to solve for 2 variables you need 2 equations, etc).

I now feel that everything about a DS question is in my mind, instead of writing down equations and , worse: doing them. Once you get your error pattern, the DS sufficiency reasoning will come almost instinctly.