[iradik]

Since one of your complaints is you don't know what problem to solve, one thing I might suggest is to try doing puzzles / challenges... They are much less daunting than trying to define a big project.

This is a great book filled with such puzzles and advice: http://www.programming-challenges.com/pg.php?page=index

Pick a problem and try to solve it. Then you can submit it to a robot judge and get immediate feedback.

There's a lot of pros here: can just do it anytime (commit 1 hr per day to working on a puzzle), can try different puzzles, learn about algorithms, get immediate feedback, and become a better coder!

Keep a git repo of all your solutions... so you can keep track of your progress.