Return two objects from lapply

Question

I have created a function which takes a little while to run (lots of crunching going on) and there are two distinct outputs that I need to return from this function. The inputs into these outputs are the same which is why I have combined them in the same function so that I don't have to crunch them twice, but the outputs are so entirely different in content and based on such entirely different calculations that there is no way to actually combine them into a one parse kinda statement. One object is tens of lines earlier than the other. But I need to return both, so I think it has to be in some type of format which mimics: store the two separate objects in a single list, lapply, then extract and rbind the two objects.

Any help on a solution to this would be appreciated - ideally not using a for loop or data.table. Dplyr solutions are fine.

Some dummy data:

df <- data.frame(ID = c(rep("A",10), rep("B", 10), rep("C", 10)),
                 subID = c(rep("U", 5),rep("V", 5),rep("W", 5),rep("X", 5),rep("Y", 5),rep("Z", 5)),
                   Val = c(1,6,3,8,6,5,2,4,7,20,4,2,3,5,7,3,2,5,7,12,5,3,7,1,6,1,34,9,5,3))

The function (again noting the function is much more complex than this, and I am calculating many more complex and unrelated things in each of the separate objects, not just the average!):

func <- function(x, df){

    temp <- filter(df, ID == x)

    average_id <- temp %>% group_by(ID) %>% summarise(avg = mean(Val))
    average_subid <- temp %>% group_by(ID, subID) %>% summarise(avg = mean(Val))

    df_list <- list(avgID=average_id, avgSubID=average_subid)

    return(df_list)

}

Presently I have computed the results using this command, but am unsure whether this is correct or how to further extract the results after the objects are stored in this list (of lists) (i.e. I get stuck here):

result <- lapply(list("A","B","C"), func, df)

The result should look like:

> average_ID
  ID avg
1  A 6.2
2  B 5.0
3  C 7.4


> average_subID
  ID subID   avg
1  A     U   4.8
2  A     V   7.6
3  B     W   4.2
4  B     X   5.8
5  C     Y   4.4
6  C     Z  10.4

I have previously used a for loop and stored the results in lists (i.e. avgListID[x] <- average_id, then binded together. But I don't think this is ideal.

Thanks in advance!

Answer 1

> average_ID <- aggregate(df$Val, by = list(df$ID), FUN = mean)
> 
> average_ID
  Group.1   x
1       A 6.2
2       B 5.0
3       C 7.4
> average_subID <- aggregate(df$Val, by = list(df$ID,df$subID), FUN = mean)
> 
> average_subID
  Group.1 Group.2    x
1       A       U  4.8
2       A       V  7.6
3       B       W  4.2
4       B       X  5.8
5       C       Y  4.4
6       C       Z 10.4

Answer 2

I realize this is a bit old, but since neither provided answer seems to have done the trick, how about this? Split the function into two, and run each within your lapply, returning a list of lists?

library(dplyr)

df <- data.frame(ID = c(rep("A",10), rep("B", 10), rep("C", 10)),
                 subID = c(rep("U", 5),rep("V", 5),rep("W", 5),rep("X", 5),rep("Y", 5),rep("Z", 5)),
                 Val = c(1,6,3,8,6,5,2,4,7,20,4,2,3,5,7,3,2,5,7,12,5,3,7,1,6,1,34,9,5,3))

subfunc1 <- function(temp){
  return(temp %>% group_by(ID) %>% summarise(avg = mean(Val)))
}

subfunc2 <- function(temp){
  return(temp %>% group_by(ID, subID) %>% summarise(avg = mean(Val)))
}

func <- function(x, df){

  temp <- filter(df, ID == x)

  df_list <- list(avgID=subfunc1(temp), avgSubID=subfunc2(temp))

  return(df_list)

}

result <- lapply(list("A","B","C"), func, df)

To get the structure/order you need, transpose the lists as explained here:

n <- length(result[[1]]) # assuming all lists in result have the same length
result <- lapply(1:n, function(i) lapply(result, "[[", i))

Answer 3

What about returning a list where each element represents the averages at a specific grouping level. For example:

library(tidyverse)

fnc = function(groups=NULL, data=df) {

  groups=as.list(groups)

  data %>% 
    group_by_(.dots=groups) %>%
    summarise(avg=mean(Val))
}

list(Avg_Overall=NULL, Avg_by_ID="ID", Avg_by_SubID=c("ID","subID")) %>% 
  map(~fnc(.x))

$Avg_Overall
# A tibble: 1 x 1
    avg
  <dbl>
1   6.2

$Avg_by_ID
# A tibble: 3 x 2
      ID   avg
  <fctr> <dbl>
1      A   6.2
2      B   5.0
3      C   7.4

$Avg_by_SubID
# A tibble: 6 x 3
# Groups:   ID [?]
      ID  subID   avg
  <fctr> <fctr> <dbl>
1      A      U   4.8
2      A      V   7.6
3      B      W   4.2
4      B      X   5.8
5      C      Y   4.4
6      C      Z  10.4

You could also just calculate the average by subID and then the average by ID can be calculated from that:

# Average by subID
avg = df %>% group_by(ID, subID) %>%
  summarise(n = n(),
            avg = mean(Val))

# Average by ID
avg %>% 
  group_by(ID) %>% 
  summarise(avg = sum(avg*n)/sum(n))

# Overall average
avg %>% 
  ungroup %>%
  summarise(avg = sum(avg*n)/sum(n))