Fitting parameter error large

1 次查看（过去 30 天）

Fei 2013-7-3

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/80954-fitting-parameter-error-large

Hello everyone,

I am trying to fit a data set with Lorentzian to extract area of the peak. For different data, the peak shape looks different. Some seem symmetric with just one peak while the others seem asymmetric with more than one peak. I used three methods to do the fitting.

First I fit all the data with a one-term Lorentzian. The area errors are low for all the peaks however the fittings look poor for the asymmetric peaks;

Then I tried to fit all the data with a two-term Lorentzian. The fittings look perfect but the area errors are horrible for all the peaks;

I also fitted separately the symmetric ones with a one-term Lorentzian while the asymmetric ones with a two-term Lorentian. The fittings to the symmetric ones look great with low area errors; the fittings to the asymmetric ones look wonderful however with still horrible area errors.

Which method do you think is the best? what strategy can I adopt to improve the fittings of the asymmetric peaks?

Thanks in advance.

Fei