quantifing goodness of fit between two data sets

Question

yuval steinberg 2021-1-6

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/710383-quantifing-goodness-of-fit-between-two-data-sets

评论： Star Strider 2021-1-7

Hi everyone,

I have two data sets, one is empirical data gathered from an experiment and the other is data from a model that is suppose to simulate the experiment. Eventually it's just two vectors. I want to quantify the goodness of fit between the model and empirical data to a single number. how can I do this?

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 2021-1-6

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/710383-quantifing-goodness-of-fit-between-two-data-sets#answer_592323

在 MATLAB Online 中打开

If ‘Q’ are your data and ‘Qest’ are the model values, one option is to compute the Coefficient of determination (

) value:

RMSE = sqrt(mean((Q -  Qest).^2));              % Compute RMSE
OLSCF = @(b) sum((Q-Qfcn(b,yez)).^2);
SStot = sum((Q - mean(Q)).^2);
SSres = OLSCF(B);
Rsq   = 1 - (SSres/SStot);                      % Compute R-squared

.

6 个评论
显示 4更早的评论隐藏 4更早的评论

Star Strider 2021-1-6

在 MATLAB Online 中打开

My pleasure!

The ‘b’ is an argument to the ‘OLSCF’ (ordinary least squares cost function) anonymous function. That function computes the sum-of-squares for a funciton that did not make it to my Answer.

A different, better and more understandable bit of code to do the same thing is:

yfit = @(b,x) exp(b(1)) .* x.^b(2);                         % Power Function
SSECF = @(b) sum((yfit(b,x) - y).^2);                       % Sum-Squared-Error Cost Function
B = fminsearch(SSECF, [1; 1]);
ypred = yfit(B,x);
SSE=sum((y-ypred).^2);
SST=sum((y-mean(y)).^2);
Rsq   = 1 - (SSE/SST);

Here, the idea is to compute the

value for the fit of this particular power function given by ‘yfit’ (where ‘x’ and ‘y’ are the data vectors) and the data, using the fminsearch function to do the nonlinear regression. Here ‘B’ is the coefficients vector comkputed by fminsearch.

Following that, the code calculates ‘ypred’ (the predicted value, corresponding to your model vector) and the actual data. The ‘SSE’ value is explained sum-of-squares, and ‘SST’ is the total sum-of-squares, as described in the Wikipedia article. .

So here, ‘ypred’ is your model, ‘y’ your data vector, and ‘Rsq’ the desired result.

My apologies for the confusion, since I should have chosen this example instead in the beginning.