Second order polynomial coefficients with one equation
12 Ansichten (letzte 30 Tage)
Ältere Kommentare anzeigen
Andreas Volden
am 30 Nov. 2014
Kommentiert: Image Analyst
am 19 Nov. 2017
Hi! I have the following equation: y=c*u+d*u^2
Known variables are y and u for a given timeseries. c and d are unknown constants. Observe that it don't exist a constant term in the equation.
I'm sure that one solution lies within least squares, but I've sort of given up without assistance. Is there another, less complex way to solve this pherhaps?
Anyone who could help progress with this problem?
Thanks!
3 Kommentare
Akzeptierte Antwort
Image Analyst
am 1 Dez. 2014
I don't see Azzi's solution here any longer so I'll just give the standard least squares solution here:
This is your model:
y = alpha1 * u + alpha2 * u^2
I don't think you can use polyfit() because there's no constant term, but you can use the standard least squares formula
alpha = inv(x' * x) * x' * y; % Get estimate of the alphas.
Where x = an N rows by 2 columns matrix.
u(1), u(1)^2
u(2), u(2)^2
u(3), u(3)^2
u(4), u(4)^2
...
u(N), u(N)^2
Now of course alpha(1) is what you called c and alpha(2) is what you called d. This is pretty much just straight out of the least squares derivation in a textbook of mine.
2 Kommentare
Weitere Antworten (1)
MariapL
am 19 Nov. 2017
hi , I am new here, looking for a solution for the same problem. I am trying to use what you said in matlab, but its not working. Could you please take a look ? Maybe you will know what I am doing wrong. So I have the same equation: y=aN^2+bN , with coefficient c already set as 0. I am typing in matlab ( my date is a time series)
x=[0 1 2 3 4 5]
y=[100 250 680 150 200 221]
y = alpha1 * x + alpha2 * x^2
alpha = inv(x' * x) * x' * y
But I will need alpha1 and alpha2 to be known in matlab in order for this to work. I dont get what should I do here to get this two coefficient.
4 Kommentare
Image Analyst
am 19 Nov. 2017
No, you don't put X into the alpha equation, you put the matrix built from X, like this:
x=[0 1 2 3 4 5]
y=[100 250 680 150 200 221]
plot(x, y, 'b*');
A = [x', x'.^2]
alpha = inv(A' * A) * A' * y'
% yFit = alpha1 * x + alpha2 * x^2
yFit = alpha(1) * x + alpha(2) * x.^2
hold on;
plot(x, yFit, 'rd-');
grid on;
legend('Training data', 'Fitted Data');
Siehe auch
Kategorien
Mehr zu Polynomials finden Sie in Help Center und File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!