Hi list, One of my co-workers found this problem with 'cor' in his code and I confirm it too (see below). He's using R 2.2.1 under Win 2K and I'm using R 2.3.0 under Win XP. ==========================================> R.Version() $platform [1] "i386-pc-mingw32" $arch [1] "i386" $os [1] "mingw32" $system [1] "i386, mingw32" $status [1] "" $major [1] "2" $minor [1] "3.0" $year [1] "2006" $month [1] "04" $day [1] "24" $`svn rev` [1] "37909" $language [1] "R" $version.string [1] "Version 2.3.0 (2006-04-24)"> data(iris) > cor(iris[1:4])Sepal.Length Sepal.Width Petal.Length Petal.Width Sepal.Length 1.0000 -0.1176 0.8718 0.8179 Sepal.Width -0.1176 1.0000 -0.4284 -0.3661 Petal.Length 0.8718 -0.4284 1.0000 0.9629 Petal.Width 0.8179 -0.3661 0.9629 1.0000> cor(iris[1:4])==1Sepal.Length Sepal.Width Petal.Length Petal.Width Sepal.Length TRUE FALSE FALSE FALSE Sepal.Width FALSE TRUE FALSE FALSE Petal.Length FALSE FALSE TRUE FALSE Petal.Width FALSE FALSE FALSE TRUE> cor(iris[1:4], iris[1:4])Sepal.Length Sepal.Width Petal.Length Petal.Width Sepal.Length 1.0000 -0.1176 0.8718 0.8179 Sepal.Width -0.1176 1.0000 -0.4284 -0.3661 Petal.Length 0.8718 -0.4284 1.0000 0.9629 Petal.Width 0.8179 -0.3661 0.9629 1.0000> cor(iris[1:4], iris[1:4])==1Sepal.Length Sepal.Width Petal.Length Petal.Width Sepal.Length TRUE FALSE FALSE FALSE Sepal.Width FALSE TRUE FALSE FALSE Petal.Length FALSE FALSE FALSE FALSE Petal.Width FALSE FALSE FALSE TRUE ========================================== The two ways of calculating correlation seem to generate the 'same' results, but the second one doesn't appear to be numerically stable (see the 3rd diagonal element of the last matrix). thanks, ...Tao _________________________________________________________________ Join the next generation of Hotmail and you could win the adventure of a lifetime
Looks like another case of the most F of all FAQs: FAQ 7.31.
See if the following makes sense to you:
pl <- iris[101:150, 3]
all.equal(cor(pl,pl), 1)
[1] TRUE
cor(pl,pl) == 1
[1] FALSE
sprintf("%1.22g", cor(pl, pl))
[1] "0.99999999999999989"
sprintf("%1.22g", pl)
[1] "6" "5.0999999999999996"
"5.9000000000000004"
etc
Peter Ehlers
Tao Shi wrote:
> Hi list,
>
> One of my co-workers found this problem with 'cor' in his code and
I confirm it too (see below). He's using R 2.2.1 under Win 2K and I'm
using R 2.3.0 under Win XP.
>
> ==========================================>
>>R.Version()
>
> $platform
> [1] "i386-pc-mingw32"
>
> $arch
> [1] "i386"
>
> $os
> [1] "mingw32"
>
> $system
> [1] "i386, mingw32"
>
> $status
> [1] ""
>
> $major
> [1] "2"
>
> $minor
> [1] "3.0"
>
> $year
> [1] "2006"
>
> $month
> [1] "04"
>
> $day
> [1] "24"
>
> $`svn rev`
> [1] "37909"
>
> $language
> [1] "R"
>
> $version.string
> [1] "Version 2.3.0 (2006-04-24)"
>
>>data(iris)
>>cor(iris[1:4])
>
> Sepal.Length Sepal.Width Petal.Length Petal.Width
> Sepal.Length 1.0000 -0.1176 0.8718 0.8179
> Sepal.Width -0.1176 1.0000 -0.4284 -0.3661
> Petal.Length 0.8718 -0.4284 1.0000 0.9629
> Petal.Width 0.8179 -0.3661 0.9629 1.0000
>
>>cor(iris[1:4])==1
>
> Sepal.Length Sepal.Width Petal.Length Petal.Width
> Sepal.Length TRUE FALSE FALSE FALSE
> Sepal.Width FALSE TRUE FALSE FALSE
> Petal.Length FALSE FALSE TRUE FALSE
> Petal.Width FALSE FALSE FALSE TRUE
>
>>cor(iris[1:4], iris[1:4])
>
> Sepal.Length Sepal.Width Petal.Length Petal.Width
> Sepal.Length 1.0000 -0.1176 0.8718 0.8179
> Sepal.Width -0.1176 1.0000 -0.4284 -0.3661
> Petal.Length 0.8718 -0.4284 1.0000 0.9629
> Petal.Width 0.8179 -0.3661 0.9629 1.0000
>
>>cor(iris[1:4], iris[1:4])==1
>
> Sepal.Length Sepal.Width Petal.Length Petal.Width
> Sepal.Length TRUE FALSE FALSE FALSE
> Sepal.Width FALSE TRUE FALSE FALSE
> Petal.Length FALSE FALSE FALSE FALSE
> Petal.Width FALSE FALSE FALSE TRUE
> ==========================================>
> The two ways of calculating correlation seem to generate the 'same'
results, but the second one doesn't appear to be numerically stable (see the
3rd diagonal element of the last matrix).
>
> thanks,
>
> ...Tao
>
> _________________________________________________________________
> Join the next generation of Hotmail and you could win the adventure of a
lifetime
>
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide!
http://www.R-project.org/posting-guide.html
--
Peter Ehlers
Chair, Division of Statistics and Actuarial Science
Department of Mathematics and Statistics
University of Calgary, 2500 University Dr. NW ph: 403-220-3936
Calgary, Alberta T2N 1N4, CANADA fax: 403-282-5150
email: ehlers at math.ucalgary.ca
Hi Peter, Thank you very much for your quick reply! I'm aware of the accuracy issue of the numbers in R. I guess the thing I was puzzled is how come the same function when used in different way produce different results (it's supposed to use the same algorithm). The only explanation for this is when cor(x) is used, the function automatically assign all the diagonal elements to 1. But when cor(x, y) is used, since there is no way to check which two vectors are the same, the function gives the 'real' results. best, ...Tao> Date: Wed, 31 May 2006 16:49:18 -0600> From: ehlers@math.ucalgary.ca> To: shitao@hotmail.com> CC: r-help@stat.math.ethz.ch> Subject: Re: [R] a problem 'cor' function> > Looks like another case of the most F of all FAQs: FAQ 7.31.> > See if the following makes sense to you:> > pl <- iris[101:150, 3]> all.equal(cor(pl,pl), 1)> [1] TRUE> cor(pl,pl) == 1> [1] FALSE> sprintf("%1.22g", cor(pl, pl))> [1] "0.99999999999999989"> sprintf("%1.22g", pl)> [1] "6" "5.0999999999999996" "5.9000000000000004"> etc> > Peter Ehlers> > Tao Shi wrote:> > > Hi list,> > > > One of my co-workers found this problem with 'cor' in his code and I confirm it too (see below). He's using R 2.2.1 under Win 2K and I'm using R 2.3.0 under Win XP.> > > > ===========================================> > > >>R.Version()> > > > $platform> > [1] "i386-pc-mingw32"> > > > $arch> > [1] "i386"> > > > $os> > [1] "mingw32"> > > > $system> > [1] "i386, mingw32"> > > > $status> > [1] ""> > > > $major> > [1] "2"> > > > $minor> > [1] "3.0"> > > > $year> > [1] "2006"> > > > $month> > [1] "04"> > > > $day> > [1] "24"> > > > $`svn rev`> > [1] "37909"> > > > $language> > [1] "R"> > > > $version.string> > [1] "Version 2.3.0 (2006-04-24)"> > > >>data(iris)> >>cor(iris[1:4])> > > > Sepal.Length Sepal.Width Petal.Length Petal.Width> > Sepal.Length 1.0000 -0.1176 0.8718 0.8179> > Sepal.Width -0.1176 1.0000 -0.4284 -0.3661> > Petal.Length 0.8718 -0.4284 1.0000 0.9629> > Petal.Width 0.8179 -0.3661 0.9629 1.0000> > > >>cor(iris[1:4])==1> > > > Sepal.Length Sepal.Width Petal.Length Petal.Width> > Sepal.Length TRUE FALSE FALSE FALSE> > Sepal.Width FALSE TRUE FALSE FALSE> > Petal.Length FALSE FALSE TRUE FALSE> > Petal.Width FALSE FALSE FALSE TRUE> > > >>cor(iris[1:4], iris[1:4])> > > > Sepal.Length Sepal.Width Petal.Length Petal.Width> > Sepal.Length 1.0000 -0.1176 0.8718 0.8179> > Sepal.Width -0.1176 1.0000 -0.4284 -0.3661> > Petal.Length 0.8718 -0.4284 1.0000 0.9629> > Petal.Width 0.8179 -0.3661 0.9629 1.0000> > > >>cor(iris[1:4], iris[1:4])==1> > > > Sepal.Length Sepal.Width Petal.Length Petal.Width> > Sepal.Length TRUE FALSE FALSE FALSE> > Sepal.Width FALSE TRUE FALSE FALSE> > Petal.Length FALSE FALSE FALSE FALSE> > Petal.Width FALSE FALSE FALSE TRUE> > ===========================================> > > > The two ways of calculating correlation seem to generate the 'same' results, but the second one doesn't appear to be numerically stable (see the 3rd diagonal element of the last matrix).> > > > thanks,> > > > ...Tao > > > > _________________________________________________________________> > Join the next generation of Hotmail and you could win the adventure of a lifetime> > > > ______________________________________________> > R-help@stat.math.ethz.ch mailing list> > https://stat.ethz.ch/mailman/listinfo/r-help> > PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html> > -- > Peter Ehlers> Chair, Division of Statistics and Actuarial Science> Department of Mathematics and Statistics> University of Calgary, 2500 University Dr. NW ph: 403-220-3936> Calgary, Alberta T2N 1N4, CANADA fax: 403-282-5150> email: ehlers@math.ucalgary.ca>_________________________________________________________________ It’s the future of Hotmail: Try Windows Live Mail beta [[alternative HTML version deleted]]
Maybe Matching Threads
- A possible too old question on significant test of correlation matrix
- saving sublist lda object with save.image()
- kernlab kpca predict
- split data, but ensure each level of the factor is represented
- unexpected behavior of trellis calls inside a user-defined function