how to plot prediction and confidence interval

Hi all,

I try to plot a prediction interval and a Confidence interval, of a linear regression fit. The prediction interval seem to be fine, but the confidence interval seems to be wrong. For the confidence interval I use ‘’ confint’’, see File.

Suggestions? Thanks a lot!!!

Matlab r2014a

[the cyclist]

Could you upload the image you get? I don't have the curve-fitting toolbox, so I can't run your code, but I'm curious about your result.

Also, can you be more specific about why you think the confidence interval is wrong? (Sometimes people have misconceptions about the relationship between these intervals.)

Answer 1

[dpb]

p = predint(fitresult,x,0.95,'observation','off');
ci = confint(fitresult,0.95);
ciup = polyval( ci(2,:),x);
cilow = polyval(ci(1,:),x);

NO! ci ARE the confidence limits; they are bounds on the coefficients themselves, hence only an upper/lower for each of the N+1 coefficients. Remove the last two lines entirely; they're a total misuse of ci

All you can plot for them would be something like

   errorbar([0 1],coeffvalues(fitresult),ci(1,:),ci(2,:),'x')

Answer 2

[piethal]

Thanks I knew there was something wrong whit my understanding of confint. Can I use

ci = predint(fitresult,x,0.95,'functional','off');

to calculate the confidence interval?

Thanks,

[dpb]

Confidence interval of what??? predint is, as the name implies and the document states, the prediction interval for a new observation.

"Confidence interval" refers to a sample statistic like the coefficients or the sample mean or the like.

[the cyclist]

You can generate a confidence interval for the fitted curve (as opposed to the coefficients of the curve). The only way I know how to do this is via bootstrap resampling. See my answer to this question for an example.