In the documentation for the seqdist() function it is noted that there is "...a specific normalization for"CHI2" and "EUCLID". See the Details section." (p.60). But in the details section there is no additional information about what normalization is applied for either of these distance measures. Thanks


This looks like an error in the documentation. Currently (TraMineR v 2.0-11.1) normalization has no effect for CHI2 and EUCLID distances. For example, you can observe with the following code that non-normalized and normalized distances are the same for a sample of 50 randomly generated sequences.

dat <- seqgen(50,12,1:3,c(.1,.5,.4))
seq <- seqdef(dat)
d <- seqdist(seq, method="EUCLID", norm="none", step=4)
d.n <- seqdist(seq, method="EUCLID", norm="auto", step=4)
sum(d - d.n != 0)

## [1] 0

May be the normalization affected the outcome before the bug fixes in version 2.0-10.

  • $\begingroup$ Thanks for clarifying - I found the same thing with my data (no effect when switching norm from "auto" to "none") but thought it best to check. $\endgroup$ – Aaron Conway May 8 '19 at 15:55

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.