[R] kSamples ad.test question

Andras Farkas motyoc@k@ @ending from y@hoo@com
Thu Aug 2 21:00:08 CEST 2018


Dear All,

once we run the following code, the results of the test will give us the expected obvious, samples are from the common distribution...


library(kSamples)

u1 <- sample(rnorm(500,10,1),20,replace = TRUE)
u2 <- sample(rnorm(500,10,1),20,replace = TRUE)
u3 <- sample(rnorm(500,10,1),20,replace = TRUE)
u4 <- sample(rnorm(500,10,1),20,replace = TRUE)
u5 <- sample(rnorm(500,10,1),20,replace = TRUE)

ad.test(u1, u2, u3,u4,u5, method = "exact", dist = FALSE, Nsim = 1000)

next, if I change "u5" to:

u5 <- sample(rnorm(500,20,1),20,replace = TRUE)

the results of the test again gives us what we expect, ie samples are not from the common distribution.... my question is: would you know of a way to be able to automatically select out or identify  "u5", the distribution that is "responsible"  for the results generated showing that the samples are not from the common distribution? 

much appreciate your help,


Andras 



More information about the R-help mailing list