How do I use arrange inside a function? - r

Question

I am trying to create a user-defined function which carries out some data transformations.
Mock data:
library(tidyverse)
set.seed(1)
sampledata_a <- data.frame(
patientid = sample(1:100),
servicetype = sample(c("service1", "service2", "service3", "service4", "service5"), 100, replace=TRUE),
date = sample(seq(as.Date('1999/01/01'), as.Date('2000/01/01'), by="day"), 100)
)
sampledata_b <- data.frame(
patientid = sample(1:100),
servicetype = sample(c("service6", "service7", "service8", "service9", "service10"), 100, replace=TRUE),
date = sample(seq(as.Date('1999/01/01'), as.Date('2000/01/01'), by="day"), 100)
)
sampledata1 <- rbind(sampledata_a, sampledata_b)
User-defined function:
get_most_recent_and_unique <- function(inputdata, groupbyvar, uniquevar, datevar) {
# first selects the most recent observation for each unique variable combination
outputdata <- inputdata %>%
distinct() %>%
arrange(groupbyvar, uniquevar, desc(datevar)) %>%
mutate(orderkey = paste0(groupbyvar, uniquevar, sep = "")) %>%
group_by(orderkey) %>%
do(head(., n=1)) %>%
ungroup() %>%
arrange(groupbyvar, desc(datevar), uniquevar)
# then tranpose from long to wide, and unite variables other than first variable into one
outputdata <- outputdata %>%
select(groupbyvar, uniquevar) %>%
group_by(groupbyvar) %>%
mutate(pos=1:n()) %>%
spread(pos, uniquevar) %>%
unite(uniquevar, -groupbyvar, sep=" / ")
return(outputdata)
}
When running the function as below:
outputdata <- get_most_recent_and_unique(sampledata1, "patientid", "servicetype", "date")
Following error message:
Error in arrange_impl(.data, dots) :
incorrect size (1) at position 1, expecting : 100
However, the code works fine when outside the user-defined function. I wonder if anyone can tell me what is wrong?
testoutputdata <- sampledata1 %>%
distinct() %>%
arrange(patientid, servicetype, desc(date)) %>%
mutate(orderkey = paste0(patientid, servicetype, sep = "")) %>%
group_by(orderkey) %>%
do(head(., n=1)) %>%
ungroup() %>%
arrange(patientid, desc(date), servicetype)
testoutputdata <- testoutputdata %>%
select(patientid, servicetype) %>%
group_by(patientid) %>%
mutate(pos=1:n()) %>%
spread(pos, servicetype) %>%
unite(servicetype, -patientid, sep=" / ")

Answer 1

Try this:
get_most_recent_and_unique <- function(inputdata, groupbyvar, uniquevar, datevar) {
groupbyvar <- enquo(groupbyvar)
uniquevar <- enquo(uniquevar)
datevar <- enquo(datevar)
# first selects the most recent observation for each unique variable combination
outputdata <- inputdata %>%
distinct() %>%
arrange(!! groupbyvar, !! uniquevar, desc(!! datevar)) %>%
mutate(orderkey := paste0(!! groupbyvar, !! uniquevar, sep = "")) %>%
group_by(orderkey) %>%
do(head(., n=1)) %>%
ungroup() %>%
arrange(!! groupbyvar, desc(!! datevar), !! uniquevar)
# then tranpose from long to wide, and unite variables other than first variable into one
outputdata <- outputdata %>%
select(!! groupbyvar, !! uniquevar) %>%
group_by(!! groupbyvar) %>%
mutate(pos=1:n()) %>%
spread(pos, !! uniquevar) %>%
unite(!! uniquevar, -!! groupbyvar, sep=" / ")
return(outputdata)
}
outputdata <- get_most_recent_and_unique(sampledata1, patientid, servicetype, date) # No quotation with arguments!
Here is the output:
patientid servicetype
<int> <chr>
1 1 service7 / service3
2 2 service10 / service1
3 3 service4 / service9
4 4 service8 / service3
5 5 service6 / service1
It seems to match your expectations when I compare them:
all.equal(outputdata, testoutputdata)
[1] TRUE
Note that you shouldn't quote the arguments when specifying the function, i.e. outputdata <- get_most_recent_and_unique(sampledata1, patientid, servicetype, date) will work while outputdata <- get_most_recent_and_unique(sampledata1, "patientid", "servicetype", "date") won't.