Excluding data on log scale

Question

Ron 2014-8-18

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/151380-excluding-data-on-log-scale

编辑： dpb 2014-8-18

Hi!

i have noisy data, and i want to fit a plot through it. i need to exclude some data points in the outliers, but i can recognize them only in the log scale plot.

is there a way to do it ?

I know that you can graphically exclude data in cftool, but it's not possible to show the data in log scale there.

Thank you

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

dpb 2014-8-18

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/151380-excluding-data-on-log-scale#answer_148992

编辑：dpb 2014-8-18

The "trick" w/ cftool is to "Print to figure" from the File Menu wherein you can use the figure editor to change axes properties. Why it's not internal to the tool is, I suppose, simply one of complexity/newer implementation doesn't yet have all the bells and whistles. I note it takes quite a long time to load already here in R2012b.

Alternatively, of course, you can use a custom fit and take log(x) on linear axes to achieve same effect albeit w/o the convenience of the scale being in original units.

2 个评论
显示无隐藏无

Ron 2014-8-18

But when i "print to figure" i can't use the exclude data option, so the problem remains :(

dpb 2014-8-18

编辑：dpb 2014-8-18

Yes, best you can do is use the other figure to identify which points you want excluded. As noted, it's an imperfect tool at best.

ADDENDUM

I'd probably approach the problem by using a log or semilog fit whichever it is that works and then compute the residuals and look for outliers numerically. A useful background article on the subject...

http://www.itl.nist.gov/div898/handbook/eda/section3/eda35h.htm

请先登录，再进行评论。