Dear R experts:
Thank you Dennis and David ...
As David indicated sorry of language and I have tried to explain what I
intend to do... I would this with Dennis's solution code:
ped <- rep(1:3, c(4, 3, 3))
> y <- rnorm(10, 8, 2)
> # This replaces all of your sample() statements, and is equivalent:
> smat <- matrix(sample(1:3, 120, replace = TRUE), ncol = 12)
> colnames(smat) <- c('M1a', 'M1b', 'M1aP1', 'M1bP2',
> 'M2a', 'M2b', 'M2aP1', 'M2bP2',
> 'M3a', 'M3b', 'M3aP1', 'M3bP2')
> mydf <- as.data.frame(cbind(ped, y, smat))
>
>
> mmat <- matrix
(c("M1a","M2a","M3a","M1b","M2b","M3b","M1aP1","M2aP1","M3aP1",
"M1bP2","M2bP2","M3bP2"), ncol = 4)
[,1] [,2] [,3] [,4]
[1,] "M1a" "M1b" "M1aP1" "M1bP2"
[2,] "M2a" "M2b" "M2aP1" "M2bP2"
[3,] "M3a" "M3b" "M3aP1" "M3bP2"
I want to compare [,1] and [,3] names of mydf (mydf[x[1]] == mydf[x[3]])
. for all three rows in the nmat. nmat is guiding me which variable I want
to pick while working on mydf. In my real dataset I have 1000 such set of
variables.
# first function
myfun <- function(x) {
x<- as.vector(x)
ot1 <- ifelse(mydf[x[1]] == mydf[x[3]], 1, -1)
ot2 <- ifelse(mydf[x[2]] == mydf[x[4]], 1, -1)
qt <- ot1 + ot2
return(qt)
}
qt <- apply(mmat, 1, myfun)
Solution of this will create a matrix with number of set of variables by
number of rows in the mydf
[,1] [,2] [,3]
[1,] 0 -2 0
[2,] -2 0 -2
[3,] 0 -2 0
[4,] 0 0 2
[5,] 0 -2 -2
[6,] -2 0 -2
[7,] -2 -2 0
[8,] -2 0 0
[9,] -2 0 2
[10,] 0 0 0
ydv <- c((y - mean(y))^2) # calculates mean of y and deviations from it for
each y values
[1] 9.5012525 0.2578341 1.6676271 6.3102202 12.8701830 9.5509480
[7] 0.8661107 3.1828185 0.9215140 1.0909813
qtd <- data.frame(ped, ydv, qt) # new data.frame with above function's
output with ped variable
ped ydv X1 X2 X3
1 1 9.5012525 0 -2 0
2 1 0.2578341 -2 0 -2
3 1 1.6676271 0 -2 0
4 1 6.3102202 0 0 2
5 2 12.8701830 0 -2 -2
6 2 9.5509480 -2 0 -2
7 2 0.8661107 -2 -2 0
8 3 3.1828185 -2 0 0
9 3 0.9215140 -2 0 2
10 3 1.0909813 0 0 0
Now I want to calculate Rt for each X1, X2, X3 (in real data world I will
have 1000 of them). The expected result of the following function should
look like 3 x 3 matrix. This is just example, I do have Ped around 200 and
X1 is around 1000.
# Rt values
Ped X1 X2 X3
1
2
3
# second function
myfun2 <- function(dataframe) {
vydv <- sum(ydv)*0.25
sumD <- sum(ydv * qt)
Rt <- vydv / sumD
return(Rt)
}
# using plyr
require(plyr)
dfsumd1 <- ddply(mydf,.(mydf$ped),myfun2)
dfsumd1
mydf$ped V1
1 1 -0.1047935
2 2 -0.1047935
3 3 -0.1047935
This is not what I want. I want ped wise Rt values for each of X variables
in above qtd matrix.
# Rt values
Ped X1 X2 X3
1
2
3
Then in I can sum Ped$X1, Ped$X2, Ped$X3. The idea is to calculated separate
Rt values for each variable group by Ped variables separately. Then add the
values.
Thank you so much for your time. Hope I had made it clear now.
Maya
>
>
[[alternative HTML version deleted]]
______________________________________________
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.
