Plot curve that is 1 standard deviation away from best fit curve ?

Question

RuiQi 2016-12-22

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/317833-plot-curve-that-is-1-standard-deviation-away-from-best-fit-curve

回答： Star Strider 2016-12-22

I have a plot that is quite noisy. I am able to fit it with a straight line using the poly fit line method with bisquare weights. https://www.mathworks.com/matlabcentral/answers/317817-how-to-use-least-squares-fit-with-bisquare-weights

I am now wondering if someone knows how I can fit a curve and maybe even plot the line that is roughly 1 standard deviation from the plotted curve. I am basically trying to get 2 lines that will encompass a large amount of the points. Maybe it could look something like that.. where the red represents the best fit curve and the green the plot that is 1 standard deviation away? The lines below are drawn in paint btw.

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

David Barry 2016-12-22

That's an interesting fit you have come up with. Not sure I would agree with it but it's difficult to tell with the Y axis scale you have plotted on.

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Star Strider 2016-12-22

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/317833-plot-curve-that-is-1-standard-deviation-away-from-best-fit-curve#answer_248202

Since you’re using polyfit, you can use the File Exchange polypredci (link) function. It defaults to 95% confidence intervals, so if you want one standard deviation, use 67% confidence intervals.