Generate data from different distributions via pmap()

Jenny Bryan 2018-05-08

Uniform[min, max] via `runif()`

CONSIDER:

runif(n, min = 0, max = 1)

Want to do this for several triples of (n, min, max).

Store each triple as a row in a data frame.

Now iterate over the rows.

library(tidyverse)

Notice how df’s variable names are same as runif’s argument names. Do this when you can!

df <- tribble(
  ~ n, ~ min, ~ max,
   1L,     0,     1,
   2L,    10,   100,
   3L,   100,  1000
)
df
#> # A tibble: 3 x 3
#>       n   min   max
#>   <int> <dbl> <dbl>
#> 1     1     0     1
#> 2     2    10   100
#> 3     3   100  1000

Set seed to make this repeatedly random.

Practice on single rows.

set.seed(123)
(x <- df[1, ])
#> # A tibble: 1 x 3
#>       n   min   max
#>   <int> <dbl> <dbl>
#> 1     1     0     1
runif(n = x$n, min = x$min, max = x$max)
#> [1] 0.2875775

x <- df[2, ]
runif(n = x$n, min = x$min, max = x$max)
#> [1] 80.94746 46.80792

x <- df[3, ]
runif(n = x$n, min = x$min, max = x$max)
#> [1] 894.7157 946.4206 141.0008

Think out loud in pseudo-code.

## x <- df[i, ]
## runif(n = x$n, min = x$min, max = x$max)

## runif(n = df$n[i], min = df$min[i], max = df$max[i])
## runif with all args from the i-th row of df

Just. Do. It. with pmap().

set.seed(123)
pmap(df, runif)
#> [[1]]
#> [1] 0.2875775
#> 
#> [[2]]
#> [1] 80.94746 46.80792
#> 
#> [[3]]
#> [1] 894.7157 946.4206 141.0008

Finessing variable and argument names

Q: What if you can’t arrange it so that variable names and arg names are same?

foofy <- tibble(
  alpha = 1:3,            ## was: n
  beta = c(0, 10, 100),   ## was: min
  gamma = c(1, 100, 1000) ## was: max
)
foofy
#> # A tibble: 3 x 3
#>   alpha  beta gamma
#>   <int> <dbl> <dbl>
#> 1     1     0     1
#> 2     2    10   100
#> 3     3   100  1000

A: Rename the variables on-the-fly, on the way in.

set.seed(123)
foofy %>%
  rename(n = alpha, min = beta, max = gamma) %>%
  pmap(runif)
#> [[1]]
#> [1] 0.2875775
#> 
#> [[2]]
#> [1] 80.94746 46.80792
#> 
#> [[3]]
#> [1] 894.7157 946.4206 141.0008

A: Write a wrapper around runif() to say how df vars <–> runif args.

## wrapper option #1:
##   ARGNAME = l$VARNAME
my_runif <- function(...) {
  l <- list(...)
  runif(n = l$alpha, min = l$beta, max = l$gamma)
}
set.seed(123)
pmap(foofy, my_runif)
#> [[1]]
#> [1] 0.2875775
#> 
#> [[2]]
#> [1] 80.94746 46.80792
#> 
#> [[3]]
#> [1] 894.7157 946.4206 141.0008

## wrapper option #2:
my_runif <- function(alpha, beta, gamma, ...) {
  runif(n = alpha, min = beta, max = gamma)
}
set.seed(123)
pmap(foofy, my_runif)
#> [[1]]
#> [1] 0.2875775
#> 
#> [[2]]
#> [1] 80.94746 46.80792
#> 
#> [[3]]
#> [1] 894.7157 946.4206 141.0008

You can use ..i to refer to input by position.

set.seed(123)
pmap(foofy, ~ runif(n = ..1, min = ..2, max = ..3))
#> [[1]]
#> [1] 0.2875775
#> 
#> [[2]]
#> [1] 80.94746 46.80792
#> 
#> [[3]]
#> [1] 894.7157 946.4206 141.0008

Use this with extreme caution. Easy to shoot yourself in the foot.

Extra variables in the data frame

What if data frame includes variables that should not be passed to .f()?

df_oops <- tibble(
  n = 1:3,
  min = c(0, 10, 100),
  max = c(1, 100, 1000),
  oops = c("please", "ignore", "me")
)
df_oops
#> # A tibble: 3 x 4
#>       n   min   max oops  
#>   <int> <dbl> <dbl> <chr> 
#> 1     1     0     1 please
#> 2     2    10   100 ignore
#> 3     3   100  1000 me

This will not work!

set.seed(123)
pmap(df_oops, runif)
#> Error in .f(n = .l[[c(1L, i)]], min = .l[[c(2L, i)]], max = .l[[c(3L, : unused argument (oops = .l[[c(4, i)]])

A: use dplyr::select() to limit the variables passed to pmap().

set.seed(123)
df_oops %>%
  select(n, min, max) %>% ## if it's easier to say what to keep
  pmap(runif)
#> [[1]]
#> [1] 0.2875775
#> 
#> [[2]]
#> [1] 80.94746 46.80792
#> 
#> [[3]]
#> [1] 894.7157 946.4206 141.0008

set.seed(123)
df_oops %>%
  select(-oops) %>%       ## if it's easier to say what to omit
  pmap(runif)
#> [[1]]
#> [1] 0.2875775
#> 
#> [[2]]
#> [1] 80.94746 46.80792
#> 
#> [[3]]
#> [1] 894.7157 946.4206 141.0008

A: Use a custom wrapper and absorb extra variables with ....

my_runif <- function(n, min, max, ...) runif(n, min, max)

set.seed(123)
pmap(df_oops, my_runif)
#> [[1]]
#> [1] 0.2875775
#> 
#> [[2]]
#> [1] 80.94746 46.80792
#> 
#> [[3]]
#> [1] 894.7157 946.4206 141.0008

Add the generated data to the data frame as a list-column

set.seed(123)
(df_aug <- df %>%
    mutate(data = pmap(., runif)))
#> # A tibble: 3 x 4
#>       n   min   max data     
#>   <int> <dbl> <dbl> <list>   
#> 1     1     0     1 <dbl [1]>
#> 2     2    10   100 <dbl [2]>
#> 3     3   100  1000 <dbl [3]>
#View(df_aug)

What about computing within a data frame, in the presence of the complications discussed above? Use list() in the place of the . placeholder above to select the target variables and, if necessary, map variable names to argument names. Thanks @hadley for sharing this trick.

How to address variable names != argument names:

foofy <- tibble(
  alpha = 1:3,            ## was: n
  beta = c(0, 10, 100),   ## was: min
  gamma = c(1, 100, 1000) ## was: max
)

set.seed(123)
foofy %>%
  mutate(data = pmap(list(n = alpha, min = beta, max = gamma), runif))
#> # A tibble: 3 x 4
#>   alpha  beta gamma data     
#>   <int> <dbl> <dbl> <list>   
#> 1     1     0     1 <dbl [1]>
#> 2     2    10   100 <dbl [2]>
#> 3     3   100  1000 <dbl [3]>

How to address presence of ‘extra variables’ with either an inclusion or exclusion mentality

df_oops <- tibble(
  n = 1:3,
  min = c(0, 10, 100),
  max = c(1, 100, 1000),
  oops = c("please", "ignore", "me")
)

set.seed(123)
df_oops %>%
  mutate(data = pmap(list(n, min, max), runif))
#> # A tibble: 3 x 5
#>       n   min   max oops   data     
#>   <int> <dbl> <dbl> <chr>  <list>   
#> 1     1     0     1 please <dbl [1]>
#> 2     2    10   100 ignore <dbl [2]>
#> 3     3   100  1000 me     <dbl [3]>

df_oops %>%
  mutate(data = pmap(select(., -oops), runif))
#> # A tibble: 3 x 5
#>       n   min   max oops   data     
#>   <int> <dbl> <dbl> <chr>  <list>   
#> 1     1     0     1 please <dbl [1]>
#> 2     2    10   100 ignore <dbl [2]>
#> 3     3   100  1000 me     <dbl [3]>

Review

What have we done?

Arranged inputs as rows in a data frame
Used pmap() to implement a loop over the rows.
Used dplyr verbs rename() and select() to manipulate data on the way into pmap().
Wrote custom wrappers around runif() to deal with:
- df var names != .f() arg names
- df vars that aren’t formal args of .f()
Demonstrated all of the above when working inside a data frame and adding generated data as a list-column

jennybc / row-oriented-workflows Public