So often this mistake is made. Are those the EXACT values of those coefficients? To me, that would seem a surprise, that they were estimated to EXACTLY 4 significant digits, when MATLAB does all of its computations in double precision, so roughly 16 significant digits. Hey. Anything is possible. Yeah, right. (There was some veiled sarcasm in there.)
Seriously, my point is, the coefficients of those curves are not exactly known. So you cannot expect an exact result from predictions made from those coefficients. And even at that, what you have is merely an approximation, a quadratic polynomial.
You need to remember these rules:
Rule 1: ALWAYS use the full coefficients, not a 4 digit approximation.
Rule 2: See rule 1.
As I said, this is a common mistake made, that you see 4 digits, so you assume that is all there is, forgetting that those lower order digits are actually important in what you must do.
