On Sun, Apr 15, 2012 at 02:52:11AM -0700, Manish Gupta wrote:
> Hi,
>
>
> In my case, your first guess is right. I need to rank classes based on
> their feature vector.
>
> 1 3 4 -2 0 class1
> 2 0 0 -3 0 class2
> 2 3 1 4 5 class3
> -4 -5 1 0 0 class4
>
> Like class1 > class3> class4> class2
>
> How can i implement it?
Hi.
The ordering may be defined in many ways depending on the purpose
of the ordering. Since i do not know this purpose, i can only
guess, what can be meaningful. Try the ordering by the mean
value. This can be done as follows.
class1<-c(1,3,4,-2,0)
class2<-c(2,0,0,-3,0)
class3<-c(2,3,1,4,5)
class4<-c(-4,-5,1,0,0)
mat <- rbind(class1, class2, class3, class4)
mat[order(rowMeans(mat), decreasing=TRUE), ]
[,1] [,2] [,3] [,4] [,5]
class3 2 3 1 4 5
class1 1 3 4 -2 0
class2 2 0 0 -3 0
class4 -4 -5 1 0 0
If the importance of the features is not equal, one can use
weigted mean. For example, as follows.
w <- c(1, 1, 4, 1, 1)
weightedMean <- (mat %*% w)/sum(w)
mat[order(weightedMean, decreasing=TRUE), ]
[,1] [,2] [,3] [,4] [,5]
class1 1 3 4 -2 0
class3 2 3 1 4 5
class2 2 0 0 -3 0
class4 -4 -5 1 0 0
Hope this helps.
Petr Savicky.
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
