如何在 R 中动态运行 function 1 到 n 次？

Question

I have a data set named master that contains survey data, structured like:

Pid state msr_01 foot_01 msr_02 foot_02 … msr_n foot_n

I want to have n data sets fetched from master , like:

out_01 contains: Pid state msr_01 foot_01 msrid
out_02 contains: Pid state msr_02 foot_02 msrid
out_n contains: Pid state msr_n foot_n msrid

The function below does this:

gen_wkds <- function (df, pno, st, col1, col2,newcol, newvalue){
  colnames <- c(pno, st, col1, col2)
  new_df <- df[, c(colnames)]
  colnames( new_df)[3] <- "Rate"
  colnames( new_df)[4] <- "Footnote"
  new_df[[newcol]] <- newvalue
  return(new_df)
}

How can I run this function dynamically from 1 to n times and generate n data sets?

Answer 1

This question deserves an answer or two, actually:

• The first answers OP's original question "How can I run this function dynamically from 1 to n times and generate n data sets?" (not recommended)
• The second answers the underlying question on reshaping from wide to long format with multiple value columns (recommended)

1. Run OP's function dynamically from 1 to n times

The OP has not provided a reproducible example, so we are using a made-up dataset

master

 Pid state msr_01 foot_01 msr_02 foot_02 msr_03 foot_03 1 1 OK 11 A1 21 B1 31 C1 2 2 OK 12 A2 22 B2 32 C2

The function gen_wkds() can be called multiple times using lapply()

lmaster <- lapply(1:3, function(x) 
  gen_wkds(master, "Pid", "state", sprintf("msr_%02i", x), sprintf("foot_%02i", x), "msrid", x))

which creates a list of dataframes

lmaster

 [[1]] Pid state Rate Footnote msrid 1 1 OK 11 A1 1 2 2 OK 12 A2 1 [[2]] Pid state Rate Footnote msrid 1 1 OK 21 B1 2 2 2 OK 22 B2 2 [[3]] Pid state Rate Footnote msrid 1 1 OK 31 C1 3 2 2 OK 32 C2 3

The list elements can be named by

names(lmaster) <- sprintf("out_%02i", seq_along(lmaster))

so lmaster becomes

$out_01 Pid state Rate Footnote msrid 1 1 OK 11 A1 1 2 2 OK 12 A2 1 $out_02 Pid state Rate Footnote msrid 1 1 OK 21 B1 2 2 2 OK 22 B2 2 $out_03 Pid state Rate Footnote msrid 1 1 OK 31 C1 3 2 2 OK 32 C2 3

Note that sprintf() is used with the %02i format specifier in order to create the names (2 digits padded with leading zero).

Normally, we would stop here because storing a bunch of datasets of the same structure in a list makes it easier to apply subsequent processing steps.

However, the OP has requested to generate n data sets . This can be achieved by

list2env(lmaster, envir = globalenv())

Again, this is not recommended as it clutters the workspace with n separate objects as can be seen here:

ls()

 [1] "gen_wkds" "lmaster" "master" "out_01" "out_02" "out_03"

(here, we only have 3 separate datasets but imagine n == 100 ...)

2. Reshaping from wide to long format with multiple value columns

Thanks to OP's explanation of the background of the question it is clear that the primary intent is to reshape the data from wide to long format. This is a common operation in data wrangling. So, several tools are available, eg:

• reshape() from base R
• pivot_longer() from the tidyr package
• melt() from the data.table package

My preferred option is melt() as it is straightforward to use, IMHO.

library(data.table)
long <- melt(setDT(master), id.vars = c("Pid", "state"), measure.vars = patterns("msr", "foot"), 
             variable.name = "msrid", value.name = c("Rate", "Footnote"))
long

 Pid state msrid Rate Footnote 1: 1 OK 1 11 A1 2: 2 OK 1 12 A2 3: 1 OK 2 21 B1 4: 2 OK 2 22 B2 5: 1 OK 3 31 C1 6: 2 OK 3 32 C2

Here, the reshaped data are kept in one date object long which makes it easier to apply subsequent processing steps programmatically. The subsets can by identified and selected by the value in the msrid column.

For the sake of completeness, long can be turned into separate objects as well by ( not recommended):

library(magrittr) # piping used for readabilty
split(long, by = "msrid") %>% 
  set_names(sprintf("out_%02i", seq_along(.))) %>% 
  list2env(envir = globalenv())

Data

master <- data.frame(
  Pid = 1:2, state = "OK", 
  msr_01 = 10 + 1:2, foot_01 = paste0("A", 1:2), 
  msr_02 = 20 + 1:2, foot_02 = paste0("B", 1:2),
  msr_03 = 30 + 1:2, foot_03 = paste0("C", 1:2)
)