R, applying a function to a subset of a data frame

Question

I watched online for a long time and did not see the answer to this specific question (I think).

The best way to explain myself would be with some code that replicates my problem. I made some temporary data:

x <- runif(100,1,2)
y <- runif(100,2,3)

z <- c(rep(1,100))
temp <- cbind(x,y,z)

temp[1:25,3] = temp[1:25,3] +2

temp <- as.data.frame(temp)

And this is what temp looks like

         x        y   z
1   1.512620 2.552271 3
2   1.133614 2.455296 3
3   1.543242 2.490120 3
4   1.047618 2.069474 3
.      .        .     .
.      .        .     .
27  1.859012 2.687665 1
28  1.231450 2.196395 1

and it continues until the end of the data frame (100 lines).

What I want to do is apply the function to the data frame, but to subsets of the data. So, for example, I want to apply the value of the function to the columns x and y for z = 3 and apply the average function to the columns x and y for, when z = 1. Thus, I get 4 values: the average value of x at z = 1 and for z = 3 and the average value of y for z = 1 and z = 3. For my actual data set, the number of rows when z = some value varies greatly.

, ; , , , for.

x <- c(unique(temp$z))

^^ z ( z = 3 z = 1).

for(i in x){
  assign(paste("newdata",i,sep=""),subset(temp[which(temp$z==i),],select=c("x","y")))
} 

, newdata1 newdata3, . newdata1 , z = 1, newdata3 z = 3.

library(gdata)

blah <-cbindX(newdata1,newdata3)

cbindX . , ( ). , , , for . , z, . z 1 50, newdata1, newdata2, newdata3.... ..

... :

summ.test <- apply(blah,2,function(x) { 
c(min(x,na.rm=TRUE),median(x,na.rm=TRUE),max(x,na.rm=TRUE),sum(!is.na(x)))})

         x         y         x         y
[1,]  1.028332  2.018162  1.012379  2.009595
[2,]  1.509049  2.504000  1.427981  2.455296
[3,]  1.992704  2.998483  1.978359  2.970695
[4,] 75.000000 75.000000 25.000000 25.000000

, , , . , : x, z = 1, y z = 1, x z = 3, y z = 3.

, : for loop , . , ?

, , - , ! .

Answer 1

> aggregate( . ~ z, data=temp, FUN=mean)
  z        x        y
1 1 1.505304 2.474642
2 3 1.533418 2.477191

, "". taht , "" , , "z".

Answer 2

library(data.table)
DT <- as.data.table(temp)
DT[, lapply(.SD, mean), by=z]
   z        x        y
1: 3 1.515801 2.309161
2: 1 1.509637 2.532575

base R:

with(temp, cbind(x=tapply(x, z, mean), y=tapply(y, z, mean)))
         x        y
1 1.509637 2.532575
3 1.515801 2.309161

PS, , set.seed(1) ;)

Answer 3

What I want to do is apply a function to a BUT data frame on subsets of data.

So you can use the function subsetor aggregate:

data = data.frame(x = runif(100), y = runif(100), z = sample(1:10, 100, replace = TRUE))

# get z > 3 AND z < 6 subset, for example
new.data = subset(data, z > 3 & z < 6) ## CAUTION: use &, not &&

# use newdata
cm = colMeans(new.data)
print(cm)

#        x         y         z 
#   0.4674450 0.5293915 4.5769231 

Hope this helps!