Question

I retrieve data from a SQL data frame in R using:

query <- "SELECT date, identifier, somevalue FROM mytable"
data <- sqlQuery(conn, query)

Which gives me:

> data
        date identifier somevalue
1 2011-01-01          1      0.50
2 2011-01-02          1      0.40
3 2011-01-01          2      0.70
4 2011-01-02          2      0.10
5 2011-01-03          2      0.25

data <- data.frame(date=c("2011-01-01","2011-01-02","2011-01-01","2011-01-02","2011-01-03"), identifier=c(1,1,2,2,2), somevalue=c(0.5,0.4,0.7,0.1,0.25))

I would like to convert this into a numeric matrix using date as the rownames and identifier as the colnames:

> output
             1    2
2011-01-01 0.5 0.70
2011-01-02 0.4 0.10
2011-01-03  NA 0.25

output <- matrix(c(0.5,0.4,NA,0.7,0.1,0.25),3)
rownames(output) <- c("2011-01-01","2011-01-02","2011-01-03")
colnames(output) <- c(1,2)

I can’t figure out how to do this. I’ve looked into reshape and also into match but I always fail due to having duplicate rownames or identifiers.

Answer 1

I generally use dcast from reshape2 (but there are copious ways of doing this):

dcast(data,
      date~identifier,
      fun.aggregate = sum,
      value.var = "somevalue",
      fill = NA_real_)
        date   1    2
1 2011-01-01 0.5 0.70
2 2011-01-02 0.4 0.10
3 2011-01-03  NA 0.25

step1 <- dcast(data,
               date~identifier,
               fun.aggregate = sum,
               value.var = "somevalue",
               fill = NA_real_)
m <- as.matrix(step1[,-1])
rownames(m) <- step1$date
> m
             1    2
2011-01-01 0.5 0.70
2011-01-02 0.4 0.10
2011-01-03  NA 0.25

The one hiccup I ran into was using the “right” NA value, which in this case was NA_real_. dcast was throwing an error if I tried either NA or NA_integer_ which doesn’t make much sense to me, but I haven’t thought about it for very long.

Edit Ok, now I get it. The NA type needs to match the type of the rest of the data apparently. I was expecting dcast to be able to convert to the appropriate NA type, but I guess not.