How do i split a vector into a data frame by date?

Question

I am trying to split this dataframe by the date and ID:

         Id       Date    Returns
1 039229109 1996-12-31  0.4739285
2 039229109 1997-01-02 -1.8867910
3 039229109 1997-01-03  0.4807711
4 056180102 1996-12-31 -4.9504940
5 056180102 1997-01-02  2.6041627
6 056180102 1997-01-03  0.0000000
7 096650106 1996-12-31 -2.0872890
8 096650106 1997-01-02 -1.8410861
9 096650106 1997-01-03  1.4807463

So that it will look like this:

            Date  039229109 056180102 096650106
    1 1996-12-31  0.4739285 -4.950494 -2.087289
    2 1997-01-02 -1.8867910  2.604163 -1.841086
    3 1997-01-03  0.4807711  0.000000  1.480746

I have tried using:

   > aggregate(data,by = list(data$Date),identity)

But that returns this:

     Group.1      Id.1      Id.2      Id.3 Date.1 Date.2 Date.3  Returns.1  Returns.2  Returns.3
1 1996-12-31 039229109 056180102 096650106   9861   9861   9861  0.4739285 -4.9504940 -2.0872890
2 1997-01-02 039229109 056180102 096650106   9863   9863   9863 -1.8867910  2.6041627 -1.8410861
3 1997-01-03 039229109 056180102 096650106   9864   9864   9864  0.4807711  0.0000000  1.4807463

I am very unfamiliar with aggregate, and feel like this should be simple, but I cannot figure out a way to do this. (I tried using reshape, but do not understand it, and could not get meaningful results.)

Thanks for your help!

EDIT: Changed and formatted the data.

Answer 1

This is really more of a re-shaping problem than an aggregating problem. That's probably why you are having a difficult time using aggregate . So if this is your sample data

data<-structure(list(Id = c("039229109", "039229109", "039229109", 
"056180102", "056180102", "056180102", "096650106", "096650106", 
"096650106", "172736100", "172736100", "172736100", "208368100", 
"208368100", "208368100"), Date = structure(c(9861, 9863, 9864, 
9861, 9863, 9864, 9861, 9863, 9864, 9861, 9863, 9864, 9861, 9863, 
9864), class = "Date"), fg.total.returnc = c(0.4739285, -1.886791, 
0.4807711, -4.950494, 2.6041627, 0, -2.087289, -1.8410861, 1.4807463, 
-0.8130074, 0.8196712, 0.8130074, -0.1253128, -0.6273508, 0.1262665
)), .Names = c("Id", "Date", "fg.total.returnc"), row.names = c("1", 
"2", "3", "4", "5", "6", "7", "8", "9", "10", "11", "12", "13", 
"14", "15"), class = "data.frame")

then you can use the base function reshape() . For example

ww <- reshape(data, timevar="Id", idvar="Date", direction="wide")
names(ww) <- gsub("fg.total.returnc.","", names(ww), fixed=T)
ww

#         Date  039229109 056180102 096650106  172736100  208368100
# 1 1996-12-31  0.4739285 -4.950494 -2.087289 -0.8130074 -0.1253128
# 2 1997-01-02 -1.8867910  2.604163 -1.841086  0.8196712 -0.6273508
# 3 1997-01-03  0.4807711  0.000000  1.480746  0.8130074  0.1262665

This is somewhat awkward because the roles of the IDs and Dates are switched in your example. I think the nicer approach is to use the reshape2 library.

library(reshape2)
dcast(data, Date~Id)

#         Date  039229109 056180102 096650106  172736100  208368100
# 1 1996-12-31  0.4739285 -4.950494 -2.087289 -0.8130074 -0.1253128
# 2 1997-01-02 -1.8867910  2.604163 -1.841086  0.8196712 -0.6273508
# 3 1997-01-03  0.4807711  0.000000  1.480746  0.8130074  0.1262665

Answer 2

A nice alternative in base R is to use xtabs :

> xtabs(fg.total.returnc ~ Date + Id, data)
            Id
Date          039229109  056180102  096650106  172736100  208368100
  1996-12-31  0.4739285 -4.9504940 -2.0872890 -0.8130074 -0.1253128
  1997-01-02 -1.8867910  2.6041627 -1.8410861  0.8196712 -0.6273508
  1997-01-03  0.4807711  0.0000000  1.4807463  0.8130074  0.1262665

The result is a matrix with class es of xtabs and table , so if you want a data.frame , be sure to wrap the above with as.data.frame.matrix and not just as.data.frame (since the latter would just put you right back where you started).