Regression of points with desired equation

Hi all,

Im trying to fit some points of established values, by using a regression made with an own equation (dependent from another parameter).

The points are defined as follow:

Xp=[4.5 5 5.5 6 ];

Yp=[0.7 0.9 0.85 1] ;

The form of the equation must be this:

Y= 2*(1 + tanh * ( ( X + a * Val2 - b ) / c ));

where Val2 is equal to 0.5.

My necessity is to define the terms a, b, c, which are unknown.

Thanks in advance

Answer 1

[Star Strider]

Try this:

Xp=[4.5  5  5.5  6 ];
Yp=[0.7   0.9  0.85   1] ;
Val2 = 0.5;
Y = @(X,a,b,c,Val2) 2*(1 + tanh( ( X + a * Val2 - b ) ./ c ));

[Prms,rsdnrm] = fminsearch(@(b) norm(Yp - Y (Xp,b(1),b(2),b(3),Val2)), rand(3,1))

figure
plot(Xp, Yp, 'p')
hold on
plot(Xp, Y(Xp,Prms(1),Prms(2),Prms(3),Val2), '-r')
hold off
grid

.

[Sergio RUGGIERI]

Thanks, it works for me!!

[Star Strider]

As always, my pleasure!