# Save error R² of an interpolation curve in variable

1 view (last 30 days)
Mepe on 12 Mar 2020
Edited: John D'Errico on 12 Mar 2020
A polynomial compensation curve with its coefficients is determined at various measuring points.
myfit=fit(x,y,'poly2');
A=myfit.p1
B=myfit.p2
C_=myfit.p3
Now i still need the error of this interpolation (R²). Can it also be read and stored in a variable?

John D'Errico on 12 Mar 2020
Edited: John D'Errico on 12 Mar 2020
You misunderstand the idea of interpolation, confusing it with approximation.
R^2 is identically 1 for an interpolation, since interpolation is defined so that it always predicts the original data. So an interpolating spline has always zero residuals in theory, therefore, R^2 == 1.
Anyway, just read the help for fit!!!!!!!
x = rand(10,1);
y = randn(size(x));
[mdl,G] = fit(x,y,'poly2')
y = randn(size(x));
[mdl,G] = fit(x,y,'poly2')
mdl =
Linear model Poly2:
mdl(x) = p1*x^2 + p2*x + p3
Coefficients (with 95% confidence bounds):
p1 = 1.399 (-3.992, 6.791)
p2 = -0.1758 (-5.699, 5.347)
p3 = -0.4521 (-1.34, 0.4359)
G =
struct with fields:
sse: 2.72060725180258
rsquare: 0.468192584870819
dfe: 7
rmse: 0.623424557447764