Do sfit and cfit objects carry the input x,y,z data? If not, why do they consume so much memory?

Question

Matt J 2022-2-28

1
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1660045-do-sfit-and-cfit-objects-carry-the-input-x-y-z-data-if-not-why-do-they-consume-so-much-memory

编辑： Matt J 2023-11-22

fittedmodel.mat

Attached is an sfit obejct from a poly33 fit, generated with the cftool app. I am wondering why it consumes so much memory (3.3MB) considering that it only contains 10 coefficients, some formula text, and perhaps an additional 20 values for the confidence bounds. Does the sfit object contain the original x,y,z data used to generate the fit? That might account for the additional memory, but if so, is there a way to retrieve that data from the object?

load fittedmodel
fittedmodel
     Linear model Poly33:
     fittedmodel(x,y) = p00 + p10*x + p01*y + p20*x^2 + p11*x*y + p02*y^2 + p30*x^3 
                    + p21*x^2*y + p12*x*y^2 + p03*y^3
     Coefficients (with 95% confidence bounds):
       p00 =     -0.4888  (-0.5129, -0.4646)
       p10 =     0.04507  (0.04389, 0.04624)
       p01 =      0.1304  (0.125, 0.1359)
       p20 =    0.005239  (0.005174, 0.005304)
       p11 =    -0.01134  (-0.01149, -0.01119)
       p02 =   -0.009979  (-0.01038, -0.009575)
       p30 =   0.0005176  (0.0005155, 0.0005198)
       p21 =   0.0002514  (0.0002483, 0.0002546)
       p12 =   0.0005639  (0.0005586, 0.0005691)
       p03 =   0.0002402  (0.0002303, 0.0002501)
whos fittedmodel
  Name             Size            Bytes  Class    Attributes

  fittedmodel      1x1              3338  sfit               

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

Matt J 2023-11-22

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1660045-do-sfit-and-cfit-objects-carry-the-input-x-y-z-data-if-not-why-do-they-consume-so-much-memory#answer_1358112

编辑：Matt J 2023-11-22

在 MATLAB Online 中打开

Never mind. I see now that the memory consumption is on ly 3.3KB, not MB. And I see from the test below that the data size does not affect the size of the fit object, so I don't think the fit object could be carrying the data around.

N=1e4;
[x,y]=deal(rand(N,1));
fobj1=fit(x,y,'poly1');
N=1e7;
[x,y]=deal(rand(N,1));
fobj2=fit(x,y,'poly1');
whos fobj1 fobj2
  Name       Size            Bytes  Class    Attributes

  fobj1      1x1               740  cfit               
  fobj2      1x1               740  cfit               

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。

Answer 2

Avadhoot 2023-11-21

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1660045-do-sfit-and-cfit-objects-carry-the-input-x-y-z-data-if-not-why-do-they-consume-so-much-memory#answer_1356937

在 MATLAB Online 中打开

Hi Matt,

I understand that you have created a “sfit” object using the “cftool” app. The object is created by the name, “fittedmodel”. As you have mentioned correctly, the large size of “fittedmodel” is due to the fact that it stores the original fitting data as the variables “x”,”y” and “z”. You can access all the attributes of the “fittedmodel” object by first loading the object into your base workspace as follows:

load fittedmodel

After executing this command, you will see all the variables defined in the "fittedmodel.mat" file in your workspace. You can access them directly from the base workspace itself.

Refer to the below documentation for more information on “sfit” objects:

https://www.mathworks.com/help/curvefit/sfit.html#description

I hope it helps.