Question

This is a follow-up question related to my previous post. Below is a more explanatory version of “what I want to do” as opposed to “how do I make this method work”.

Below is code that produces a “master” database, from which, I extract elements for further use in other functions. I routinely extract elements of data based on the value of a group identification number.

• Objective: I would like to be able to “wrap” the specifications that vary (like the name of the output dataframe and the groups selected) into a function that could be called.

---

#####  generating data for example

set.seed(271828)

n.elements <- c(10,10,12,14,16,18)
group.number <- rep(1001:1006, n.elements)
element.id <- c(
    seq(1,n.elements[1], 1),
    seq(1,n.elements[2], 1),
    seq(1,n.elements[3], 1), 
    seq(1,n.elements[4], 1),
    seq(1,n.elements[5], 1),
    seq(1,n.elements[6], 1) ) 

x1 <- round(rnorm(length(group.number),45, 12), digits=0)
x2 <- round(rbeta(length(group.number),2,4), digits = 2)

data.base <- as.data.frame(cbind(group.number, element.id, x1, x2))
data.base

#####  data.base is representative of the large database 


#####  suppose I need to pull a set together made up of groups: 
#####  1003, 1004, and 1001 


groups.set.1 <- as.data.frame(c(1003, 1004, 1001))
bank.names <- c("group.number")
colnames(groups.set.1) <- bank.names
set.sort <- matrix(seq(1,nrow(groups.set.1),1)) 
sort.set.1 <- cbind(groups.set.1, set.sort)

set.1 <- as.data.frame(merge(sort.set.1, data.base, 
by="group.number", all.x=TRUE))

#####  this is how the dataset needs to be ordered for further use
set.1 <- set.1[order(set.1$set.sort, set.1$element.id ), ]
row.names(set.1) <- seq(nrow(set.1))

---

EDIT: Suppose I wanted to carry out the same task to produce set.2, where set.2 is made up of groups: 1005, 1006, and 1002. I could just copy the above code, and make the relevant changes. However, I would like to know if it is possible to specify a function so that I can pass the necessary changes to it, and have it produce the output dataframe as desired. Perhaps having a function called group.extract, where I could specify something like the following:

groups.2 <- c(1005, 1006, 1002)
group.extract(set.2, groups.2)

Based on the comments provided, it seems like a list is the way to go, and have the function call the list, where the list elements can vary.

Answer 1

I’d write this function using match, as follows. Here I’ve hard-coded the names of the columns of the input data frame to use for matching and sorting; those could also be added as optional inputs. The column order of the output is slightly different from yours but that could be easily changed as well.

getset <- function(g, d=data.base) {
  d$set.sort <- match(d$group.number, g)
  d <- d[!is.na(d$set.sort),]
  d <- d[order(d$set.sort, d$element.id),]
  rownames(d) <- NULL
  d
}

You’d use it almost exactly like you propose:

> set.1 <- getset(c(1003, 1004, 1001))
> head(set.1)
  group.number element.id x1   x2 set.sort
1         1003          1 60 0.32        1
2         1003          2 28 0.18        1
3         1003          3 42 0.47        1
4         1003          4 43 0.08        1
5         1003          5 45 0.31        1
6         1003          6 27 0.48        1

Though if you have multiple groups to get, putting them in a list and using lapply would be the way to go.

> groups <- list(group1=c(1003, 1004, 1001), group2=c(1005,1006,1002))
> sets <- lapply(groups, getset)
> lapply(sets, head)
$group1
  group.number element.id x1   x2 set.sort
1         1003          1 60 0.32        1
2         1003          2 28 0.18        1
3         1003          3 42 0.47        1
4         1003          4 43 0.08        1
5         1003          5 45 0.31        1
6         1003          6 27 0.48        1

$group2
  group.number element.id x1   x2 set.sort
1         1005          1 27 0.20        1
2         1005          2 51 0.48        1
3         1005          3 49 0.43        1
4         1005          4 48 0.20        1
5         1005          5 33 0.37        1
6         1005          6 41 0.50        1