Help with data-fitting using lsqcurvefit and MultiStart

Question

Nik 2017-7-31

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/350997-help-with-data-fitting-using-lsqcurvefit-and-multistart

回答： Chris Perkins 2017-8-9

I currently have a GUI that takes plots the intensity of experimental videos. The data follows a logistic curve for the most part. I'm trying to find two parameters using lsqcurvefit, and this works for the majority of the videos. However, certain plots have curves that are significantly worse than other similar videos. Would using MultiStart possibly help with the fit? So far the results are almost identical to the function without MultiStart. Do you have any recommendations about how I could be using MultiStart?

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

Chris Perkins 2017-8-9

1
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/350997-help-with-data-fitting-using-lsqcurvefit-and-multistart#answer_277440

"MultiStart" can help improve results. It works to find a global minimum, rather than a local minimum, so when "lsqcurvefit" finds a local minimum then "MultiStart" certainly could improve results - how much depends on how good of a local minimum "lsqcurvefit" found.

To learn more about how to integrate "MultiStart" into "lsqcurvefit", see the following article: https://www.mathworks.com/help/gads/multistart-using-lsqcurvefit-or-lsqnonlin.html

The function "GlobalSearch" is somewhat similar to "MultiStart", and depending on your application you may want to look into "GlobalSearch" as well. See the following article for a description of both "MultiStart" and "GlobalSearch", including how they differ: https://www.mathworks.com/help/gads/how-globalsearch-and-multistart-work.html