Curve Fitting Toolbox error before starting

Hi, I'm trying to use the Curve fitting Toolbox but I've got an error when I type the command cftool. I think is because my matlab is old and doesn't support the ~ operator, because when I go to the error points to it in the function code. But I'm not such experienced user and I don't know how to solve the problem. Changing it by 'trash' didn't help.

I'll very much appreciate any suggestion. Thank you G.

[Jiro Doke]

That doesn't make sense. Are you sure you installed the same version of Curve Fitting Toolbox as the version of MATLAB? Then that kind of error should not happen.

The only thing I can think of is that you have an old version of MATLAB, and you're trying to run functions from a newer version of Curve Fitting Toolbox. You should install the version of the toolbox for your MATLAB. You can't just copy (or run) any version. Each toolbox is meant to run on the same version of MATLAB.

Answer 1

[Andreas Goser]

In order to find out the toolbox is installed correctly:

ver

should return MATLAB as well as Curve Fitting toolbox with the same release, e.g.:

MATLAB                                                Version 7.11       (R2010b)
Curve Fitting Toolbox                                 Version 3.0        (R2010b)

And also verify correct licensing:

license('test','Curve_Fitting_Toolbox')

This should return 1.

Answer 2

[Guisella Gacitúa]

Yes, that's right.. they are different versions.. Then, the question should be how to solve the problem without using the fitting curve toolbox. What I need is to interpolate some points and to obtain a parameter that gives some confidence interval that can be used to calculate the propagation of error in some further estimations using this data. The error should be quite small in my measured points, so the further I go from the known points I should have a higher uncertainty. Thanks.

[Walter Roberson]

You need to give us an idea of the complexity of the curve. It would help if you could post a plot of the curve. Upload the plot to your favorite image sharing system and post the link to it.

[Guisella Gacitúa]

I posted a figure with the data:

https://picasaweb.google.com/107006156555720597966/Mar21201102#

I need the uncertainty range to evaluate the area from 45 to 80 meters. I use this depth information for further calculations and the linear assumption yields in high uncertainties here.

[Walter Roberson]

I think that data will be very difficult to fit meaningfully. It is far too jumpy over the target area to assign curve to it that would have enough confidence to improve beyond using trapz.

The only hope that I can see is if you have some kind of known physical model that explains the kinds of values you see, including reasoning for the jumps: with a physical model in hand, it _might_ be possible to make a better calculation.

[Guisella Gacitúa]

Thank you for your quick answer.

The thing is that I would only like to justify the "strange values" that I obtain in my further use of this data in the jumpy area, due to the linear interpolation. The fact is that the known points are quite accurate but in between there will be higher uncertainty that is propagated in the calculations afterwards. Therefore, I wanted to get this error as a reference.

I don't know if this still make sense to you..

[Andreas Goser]

Independent on whether you can resolve this without the toolbox. what is so diffucult to install the right versions?

[Guisella Gacitúa]

That was yesterday, sorry I didn't mention that. I have the toolbox working now

[Walter Roberson]

Okay, so what you have is known to be fairly precise where you have it -- but is there a good model for the transitions between the locations? If not then there isn't anything that can be done.

[Guisella Gacitúa]

No, there is not model for that. These are direct measurements of natural soil depth, it can be randomly variable in between the known points, but the variability is more pronunced only in this particular section of about 40 meters, where I'm interested on giving the uncertainty in the estimations, maybe based in the slope difference between consecutive points.