are Computer Analysis worth the bytes they are printed on?

I did several computer analysis of my games and noticed that basically they aren't worth squat..  Several times in the games I get told Inacuracy and then 3 moves later when another Incacuracy is given it says I went from a winning position to equal.  So two things, one if it was inacurate how was it winning with the next two moves and Also, and this is big.. the lines it suggests as being right often indicate the other side as ahead.  So how can that be?  It also chides me on many of my game winning tactics.  I do like the value rating but have a really hard time seeing how the suggested lines are relevant given I know my opponent would not happily follow along..  I get much more out of studying gm and really amatuer (1900-2200) rated players..  Are the computer analysis how a computer engine would play or are they sampled more from known GM positions?

They are not based on master games at all, just standard computer brute force calculation.

Good for tactical oversights, I wouldn't place much stock in the 'inaccuracies'.