How do I fit an exponential curve fit function using fminsearch. Program runs, but result is incorrect.

Question

carenar 2015-6-19

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/224575-how-do-i-fit-an-exponential-curve-fit-function-using-fminsearch-program-runs-but-result-is-incorre

评论： Star Strider 2015-6-22

Attached is my code

yfirst=[.8967e07 1.6294e07 2.6587e07 3.2537e07 3.6136e07 3.8419e07 3.9706e07]'; 
xfirst=[1,2,4,6,8,10,15];
[estimates, model] = myfun(xfirst,yfirst);
[sse, FittedCurve] = model(estimates);
semilogx(xfirst,FittedCurve,'-g*'); hold on;
semilogx(xfirst,yfirst,'-rd');hold off;
 xlim([ 0 50 ]);
%%%%FUNCTION FILE 
function [estimates, model] = myfun(xdata, ydata)
% Call fminsearch with guessed starting point.
% start_point =[2.23e7;.005];
model = @expfun;
estimates = fminsearch(model,start_point);
    function [sse, FittedCurve] = expfun(params)
        A=params(1)
        lambda=params(2)
        FittedCurve =(A .* exp(lambda * xdata));
        ErrorVector = FittedCurve - ydata;
        sse = sum(ErrorVector .^ 2);
    end
end

I get a large error and the curve does not match when plotted together. Thanks

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Star Strider 2015-6-19

2
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/224575-how-do-i-fit-an-exponential-curve-fit-function-using-fminsearch-program-runs-but-result-is-incorre#answer_183327

在 MATLAB Online 中打开

There are several problems with your code.

This works:

yfirst=[.8967e07 1.6294e07 2.6587e07 3.2537e07 3.6136e07 3.8419e07 3.9706e07]'; 
xfirst=[1,2,4,6,8,10,15]';
expfun = @(b,xdata) b(1) -b(2) .* exp(b(3) .* xdata);             % Objective Funciton
SSECF = @(b) sum((yfirst - expfun(b,xfirst)).^2);           % Sum-Squared-Error Cost Function
start_point =[4E+7; 2.23e7; -.005];
[B, SSE] = fminsearch(SSECF, start_point);
figure(1)
plot(xfirst, yfirst, 'bp')
hold on
plot(xfirst, expfun(B,xfirst), '-r')
hold off
grid
text(5.2, 1.75E+7, sprintf('f(x) = %9.2E - %9.2E\\cdote^{%9.2E\\cdotx}', B))

2 个评论
显示无隐藏无

carenar 2015-6-22

Do necessarily need the third constant term in the front?

Star Strider 2015-6-22

In this instance (that is with your data), you do. It is an asymptotically-increasing exponential, so you have to have parameters for the asymptote, amplitude, and exponential rate. The integrated differential equation for the process that created your data would require values for all three parameters. (With a simple decaying exponential, you would only need parameters for the initial value and exponential rate.)

请先登录，再进行评论。