Re: Idempotence and "Replication Insensitivity" are equivalent ?
Date: 25 Sep 2006 12:42:06 -0700
> vc ha scritto:
> Quantiles are invariant to monotonic transformations.
> > So what ?
> What? You should decide.
> You have just pointed out that the median is not invariant wrt general
> monotonic transformations (as the AVG) (you showed that using f=log),
Earlier, I said that the sample median regarded as an interval (50 percentile) *is* invariant under monotone transformations. Such treatment is customary when monotone transformations matter. When they do not, one can use a single value, for convenience, obtained as the average of two endpoints. That value of course is *not* invariant under monotonic transformations.
> and now you are saying that "Quantiles are invariant to monotonic
> transformations". Do you realize these are conflicting statements?
> I give up to understand you.
Well, I cannot help with that..
> Do not you think that before sarcastically suggesting other people (who
> you do not know) to read some book should make sure you have
> understanding of what you are talking about?
Received on Mon Sep 25 2006 - 21:42:06 CEST