Re: [R] Problems with ks.test()

(I'm replying to your original post because your follow-up omits the context.) The K-S test is designed for continuous distributions. You have far too many zeros in your data to get anything reasonable out of the test. For your data, the K-S statistic is the difference in the (e)cdfs at zero.

Re: [R] Problems with ks.test()

Hi, I used a ks-function of another library (kstwo() of numerical recipes, a mathematics book) to test it for myself and the same happens there - I cannot understand why this observation happens? I hope someone can 'enlighten' me. -- View this message in context:

Re: [R] Problems with ks.test()

data and the CDF to the function, don't worry about midpoints. -Original Message- From: r-help-boun...@r-project.org [mailto:r-help-boun...@r-project.org] On Behalf Of Jochen1980 Sent: Friday, July 29, 2011 3:08 AM To: r-help@r-project.org Subject: [R] Problems with ks.test() Hi, I got

[R] Problems with ks.test()

Hi, I got two data point vectors. Now I want to make a ks.test(). I you print both vectors you will see, that they fit pretty fine. Here is a picture: http://www.jochen-bauer.net/downloads/kstest-r-help-list-plot.png As you can see there is one histogram and moreover there is the gumbel