I don't know what the mistakes are called, but the last one is a typical one, i.e. not analyze a line fully to quiescence, or maybe evaluate it incorrectly. What you describe reminds me of my latest game.

That game I lost partly because I wanted to play a move but couldn't make it work. Therefore, I chose another move which looked like it improved the initial variation - by reversing the move order. 

However, there were two main problems with my chosen move:
Firstly, the move order was critical. Secondly, the initial thought was correct, but I stopped calculating the variation too early. I should have spent more time on it, really because the move I wanted to play felt correct and was very thematic.

Also, there were many different options for each side. Calculating a couple of moves ahead correctly was tricky. But that is normally much more important than to calculate a few long variations well, but then miss a critical early improvement for your opponent.

As for the training, I can't really tell. I will do tactics training regularly, with emphasis not on 'One side to play and win', but to calculate efficiently and correctly in many different positions. Another thing on my list is to go through well annotated games with the same focus.