Double sigmoid fit

Hi, I have bell-shaped data that I'd like to fit. Publications with related data use a "fit to the Boltzman equation." I imagine that this means a double sigmoid fit.

The curve fitting tool does not have a related option, any ideas would be greatly appreciated.

Thanks! Stefan

[Sarah Wait Zaranek]

http://www.mathworks.com/matlabcentral/newsreader/view_thread/169459

This link should help you out.

[Stefan]

Thanks for the tip. I was looking at that thread before I posted; unfortunately, I'm not able to move forward from the answers on that page.

One of the replies states that "the cftool GUI or the fit function should be able to do this. Also the nlinfit function in the Statistics Toolbox, the lsqcurvefit function in the Optimization Toolbox, and others." But it doesn't say how to do use these tools to achieve a sigmoid fit.

For example, the reply states that the cftool should be able to, however there is no sigmoid option in curve fit.

[Andrew Newell]

See also http://www.mathworks.com/help/toolbox/stats/bq_676m-2.html#bq_676m-39.

Answer 1

[Andrew Newell]

Here is an example of how you could use cfittool to create a sigmoid fit. First, enter these commands in the Command Window.

x = -5:.05:5;
y = tanh(x);
cfittool

In the Curve Fitting Tool, do the following:

1. Click on the Data... button and import x and y.
2. Click on Fitting... and then New fit.
3. Under Type of fit, choose CustomEquations.
4. In the New Custom Equation window, click on GeneralEquations.
5. In the Equation: line after the = sign, type in a*exp(-b*x)+c, and click OK.
6. Finally, click Apply to see a nicefit.