How can I modify this code so it gives the value 0 for c. It should only provide ax^2+bx

2 views (last 30 days)
function [coefficients, MSE] = pnnnfit(x, y, order)
coefficients = polyfit(x, y, order)
y_fitted = polyval(coefficients, x)
squaredError = (y - y_fitted) .^ 2
MSE = sum(squaredError)
end
%x=[0 .5 1 1.5 2 2.5 3 3.5 4 4.5 5];
% y=[0 -5.27 -8.10 -8.46 -6.38 -1.84 5.15 14.59 26.48 40.83 57.63];
ans =
4.9069 -13.0140 0.0265
  1 Comment
Walter Roberson
Walter Roberson on 4 Dec 2020
If you mean that a^2+b*x should exactly equal y, then that is not the case.
The best degree 2 fit is 61067/12450 * x^2 - 17802227/1369500*x + 10357/136950000

Sign in to comment.

Answers (1)

Asad (Mehrzad) Khoddam
Asad (Mehrzad) Khoddam on 4 Dec 2020
%x=[0 .5 1 1.5 2 2.5 3 3.5 4 4.5 5];
%y=[0 -5.27 -8.10 -8.46 -6.38 -1.84 5.15 14.59 26.48 40.83 57.63];
function [coefficients] = pnnnfit(x, y)
coefficients=[sum(x.^4) sum(x.^3) ; sum(x.^3) sum(x.^2)]\[sum(y.*x.^2) sum(y.*x)]';
% a = coefficients(1)
% b = coefficients(2)
% you need to add more code to find MSE but it is simple
plot(x,y,x,coefficients(1)*x.^2+coefficients(2)*x)
end
  2 Comments
Asad (Mehrzad) Khoddam
Asad (Mehrzad) Khoddam on 4 Dec 2020
Edited: Asad (Mehrzad) Khoddam on 4 Dec 2020
Yes, you are right. I used the actual mathematical equations for the least squares method. Matlab has a simplified '\' operator for solving equations and the least square method

Sign in to comment.

Categories

Find more on Mathematics in Help Center and File Exchange

Tags

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!