Nonlinear fit to data with errors in both coordinates

2 次查看（过去 30 天）

Niles Martinsen 2012-9-21

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/48781-nonlinear-fit-to-data-with-errors-in-both-coordinates

I have a set of data (x, y), where there are measurement errors on both coordinates. These errors are non-constant, and I wish to fit a nonlinear function to the data.

By using nlinfit I am able to fit my function to the data including the errors on y. But is there a way for me to include the ones in x as well?

I would be very happy to get a push in the right direction. I couldn't seem ti find an answer using the "search".

Thanks in advance.

Niles.

2 个评论
显示无隐藏无

Star Strider 2012-9-21

This is known as a Total Least Squares problem. There appear to be a number of File Exchange contributions on the Total Least Squares Method, but I haven't looked at them to see if any apply to nonlinear least squares. I imagine that the same techniques would apply, although I have never had occasion to use it in a nonlinear regression context.

Niles Martinsen 2012-9-22