[prev in list] [next in list] [prev in thread] [next in thread] 

List:       r-sig-epi
Subject:    Re: [R-sig-Epi] R-sig-Epi Digest, Vol 39, Issue 1
From:       Pedro Emmanuel Alvarenga Americano do Brasil <emmanuel.brasil () gmail ! com>
Date:       2010-02-02 10:33:28
Message-ID: a37cfb5a1002020233o672655dctb3cadd5f23f3228a () mail ! gmail ! com
[Download RAW message or body]

Sorry for coming late in the discussion...

I got many reply on reply on reply so im not so shure if the issue is
solved. Regarding the differences in senSpec and dignosis functioins, as far
as I remember, sensSpec gives the confidence intervals with normal
distributions while diagnosis gives with with binomial distributions which
could be an advantage while the data is highly skewed and normal
distribution may give confidence limits outside 0 1.

In a local R list I have seen people rporting problems with diagnosis
function and it always seems to be related varibles number of levels. Try
converting to characters...

I have a sugestion that may workaround:

t<-table(test,gold)
t
diagnosisI(t[2,2],t[1,2],t[2,1],t[1,1])

# Check carefully if TP FP TN and FN are correctly in the right place in the
table!

should be like:

TN  FP
FN  TP

Kind regards,

Abra=E7o forte e que a for=E7a esteja com voc=EA,

Dr. Pedro Emmanuel A. A. do Brasil
Instituto de Pesquisa Cl=EDnica Evandro Chagas
Funda=E7=E3o Oswaldo Cruz
Rio de Janeiro - Brasil

	[[alternative HTML version deleted]]



_______________________________________________
R-sig-Epi@stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-epi


[prev in list] [next in list] [prev in thread] [next in thread]