Performance curve

2 posts / 0 new
Last post

Hi All,
When i am calculating curve coefficients by enetring raw data
points for performance curve i am getting different values of curve
coefficient a,b,c,d,e,f each time where as my value of x,y, z is same why
is this so

if i have z= a+bx+cx2+dy+ey2+fxy then my a , b,c,d,e,f,y must represent
unique solution each time when my six values of z,x,y are same

Regards
Sambhav

sambhav tiwari's picture
Offline
Joined: 2011-09-30
Reputation: 0

Hi Sambhav,

What you?re describing has me concerned? for simplicity, consider the following 2D math question:

If I provide one point only: (x=1, y=2), and task you to determine the coefficients for a line intersecting that point using the form y = Ax+ B, how many correct answers are there? This question is of course flawed!

If you mean to have equest determine a bi-quadratic curve that always returns the same result independent of variables x and y, and to do so using raw data point regressions, you would need to furnish more than a repeated single raw data point for the exact same reasons.

That said, it would be an easier problem to select the curve coefficients, first looking at the basic form and figure out what values to choose for a,b,c,d,e,f if you mean for any value of x and y to cancel out: z = a + (0)x + (0)x^2+ (0)y + (0)y^2 + (0)xy. In that scenario, the coefficient ?a? as you define it would always be returned regardless of the variables x and y.

Extra advice ? you haven?t been specific to why you?re doing this, but be very sure you understand how the curve?s result (z) is used if you intend to tread the path of ?dumbing down? the curves. Attached discussion from a few months back is a good case in point and would be good review as a case in point.

Best of luck!

~Nick

[cid:489575314 at 22072009-0ABB]

NICK CATON, P.E.

Nick-Caton's picture
Offline
Joined: 2011-09-30
Reputation: 805