Problem - Optimization nonlinaer function

Hello.

I need help, to find p1, p2 and p3 to get the best Pearson correlation between x and y.

My function iz:

x= A*p1 * exp(B*p2)*exp (C*p3)

This is my code:

clear;

close all;

clc;

load('myData_all.mat');

%define A B C and y

y1 = MOS;

A1 = Metr_all;

B1 = QInitReb_all;

C1 = QBitRate_all;

%Due to different scales, rescale all data to have mean 0 and standard deviation 1

[y,ymean,ydev] = zscore(y1);

A = zscore(A1);

B = zscore(B1);

C = zscore(C1);

%define initial values for p

p0 = [0.5, 0.5, 0.5];

%Define the objective function as an anonymous function.

f = @(p)myParFun(p,A,B,C,y);

%Optimize for p

[p,fval] = fminunc(f,p0)

%Calculate x using calculated p

[r,x] = myParFun(p,A,B,C,y);

function [r,x] = myParFun(p,A,B,C,y)

x = A*p(1) * exp( B*p(2)) *exp( C*p(3));

r = myPearsonCoeff(x,y);

%since optimazation toolbox minimizes but we want to maximize

r = -r;

end

Answer 1

[Torsten]

function r = myParFun(p,A,B,C,y)
  x = A*p(1).* exp( B*p(2)) .*exp( C*p(3)); 
  r = myPearsonCoeff(x,y);
  %since optimization toolbox minimizes but we want to maximize
  r = -r;      
end

[Milica Arsenovic]

Thank you!