How can I estimate confidence bands for kernel density estimation?

Hi everyone

I have a series of residuals that are supposed to be standard normally distributed. By using the ksdensity function, I obtain a smooth curve that indeed looks like a Gaussian. However, I want to construct confidence intervals (95%) around this density plot. Is there a built-in way how to achieve that? I also tried the Distribution fitting tool, which yielded a density plot but no confidence intervals.

Thank you in advance!

Answer 1

[José-Luis]

You could use fitdist(), which gives you the bounds you are looking for.

[mr]

But how exactly do I have to proceed? Sample a sufficient number of observations and plot a histogram that depicts the pdf of the top 95% observations?