Regarding the stonewall game:
White played the opening poorly and after 10. Nb3?! (unclear what that knight is doing here), black has a very comfortable position. However 10....a6 is just weakening the queenside without any need.
At this point, you really do want to play something active that will eventually free your cramped position. Both 10... b6 or 10....e5 look like decent options and should free your position.
Your idea with Ng4 (and certainly the knight sacrifice) looks overly optimistic to me. You are not going to mate on that side of the board easily against normal defense.
The Scotch game. Another inaccurate piece sacrifice. At least I would have expected Bxb2 to develop a piece.
After that plenty of tactical blunders from both sides. A computer analysis should be able to point these out. I focus on other mistakes that are hard to see with computers. If you are attacking, generally (and particularly if you are material down) you want to avoid simplifications and exchange of pieces. Moves like 12. Nxd4 or 18. Ne6 should therefore be checked carefully.
By far the most normal of these three games. No attack going on. You should play this positionally and be concerned with not losing pawns and with positioning your pieces actively.
Ne5 looks pseudoactive: the knight looks good there for a moment, but does not really threaten anything and will be challenged soon. Potentially more serious is that you exchange your light squared bishop and stay with the worse dark squared bishop.
Qa4 would win a piece.
After 16. 0-0-0 0-0-0 would be a good time to assess the position and come up with a new plan. You are a pawn down and should at least try to position your pieces as active as possible. However within the next couple of moves, you position your pieces just to protect your own pawns, put your bishop on the poor square f4 (Be3 might have been better) and give black the open d file for the rook. Not surprisingly you lose more pawns and are just lucky at the end for black blundering a piece.
Always difficult to give an assessment based on a couple of games only. In your case I would recommend to be less optimistic in your attacks/sacrifices. Even if you end up winning now, against stronger opponents I doubt it works all that often.
Something to look out for could be piece activity and how to improve this. See what good/weak bishops are and how to get/prevent them, etc.