Question

I have data that I would like to compute some statistics with.
The data is organized in such a way that I have a value corresponding to each 3-element tuple
Something like

(P1,M1,R1,V1)
(P1,M1,R2,V2)
(P1,M2,R1,V1)
...

here P1, M1, and R1 are not numeric but V1 and V2 are.
Right now I have the data in a csv file, x2.cvs as follows:

P,M,R,V
P1,M1,R1,V1
P1,M1,R2,V2
...

I read the data using

d = read.table("x2.csv", sep=",", header=TRUE)

but after that I don’t know what to do to process the data.

I would like to start by computing simple information like:
what is the average for each element of P (so the average would be over all elements
of M and R), or for each pair of elements of {P,M} (so the average here would be over the elements of R.

Next I would like to do a little bit more complicated things like compute how many elements of P1 are bigger than some specified value.

Answer 1

Here’s a start, with data.table, plyr and base functions, there are so many other ways…

First, some example data…

dput(examp)
structure(list(P = structure(c(1L, 1L, 1L, 2L), .Label = c("P1", 
"P2"), class = "factor"), M = structure(c(1L, 1L, 2L, 2L), .Label = c("M1", 
"M2"), class = "factor"), R = structure(c(1L, 2L, 1L, 1L), .Label = c("R1", 
"R2"), class = "factor"), V = c(23, 49, 24, 29)), .Names = c("P", 
"M", "R", "V"), row.names = c(NA, -4L), class = "data.frame")
#
# to give something like what you have...
#
examp
   P  M  R  V
1 P1 M1 R1 23
2 P1 M1 R2 49
3 P1 M2 R1 24
4 P2 M2 R1 29

Here is one way using data.table. If your data object is very big, you’ll find the data.table package to be very fast, documentation is also excellent: http://datatable.r-forge.r-project.org/datatable-intro.pdf

# What is the average of each element of P?
library(data.table)
examp.dt <- data.table(examp)
setkey(examp.dt,P)
examp.dt[,mean(V),by=P]
      P V1
[1,] P1 32
[2,] P2 29
#

And another using plyr

# What is the average of each element of P?
library(plyr)
ddply(examp, "P", function(df)mean(df$V))
P V1
1 P1 32
2 P2 29

And another using base R

# What is the average of each element of P?
# for example using the by() function,  tapply() would be similar
with(examp, by(examp, P, mean))
P: P1
P  M  R  V 
NA NA NA 32 
------------------------------------------------- 
P: P2
P  M  R  V 
NA NA NA 29
#
# What is the average of each element of R? 
with(examp, by(examp, R, mean))
R: R1
P        M        R        V 
NA       NA       NA 25.33333 
----------------------------------------------
R: R2
P  M  R  V 
NA NA NA 49 
#
# the same, using tapply
with(examp, tapply(V, R, mean)
  R1       R2 
25.33333 49.00000

And for your last question, how many elements of P1 are bigger than some specified value, we can use subset like so:

# how many elements of P1 are greater than 20?
nrow(subset(examp, examp$P=="P1" & examp$V>20))
[1] 3

Or even just [ for the same result with less typing:

nrow(examp[examp$P=="P1" & examp$V>20,])
[1] 3