A challenge tweeted by Hilary Parker, paraphrased:
How do you sample from groups, with a different sample size for each group?
Illustrated with the iris data.
Species = groups.Species with 3 different sample sizes.How fits the template:
DRAW A SAMPLE for each PAIR OF (SPECIES DATA, SPECIES SAMPLE SIZE)
How to prepare the data? I need a data frame with
SpeciesSpecies-specific sample sizesSpecies-specific data frames. A list-column!We need a nested data frame.
suppressMessages(library(dplyr))
library(purrr)
library(tidyr)
set.seed(4561)
(nested_iris <- iris %>%
group_by(Species) %>% # prep for work by Species
nest() %>% # --> one row per Species
ungroup() %>%
mutate(n = c(2, 5, 3))) # add sample sizes
#> # A tibble: 3 x 3
#> Species data n
#> <fct> <list<df[,4]>> <dbl>
#> 1 setosa [50 × 4] 2
#> 2 versicolor [50 × 4] 5
#> 3 virginica [50 × 4] 3Draw the samples.
purrr::map2() is good since we want to operate on 2 things (data = DATA FOR ONE SPECIES, n = SAMPLE SIZE).data = DATA FOR ONE SPECIES and n = SAMPLE SIZE as variables in our data frame.dplyr::sample_n(tbl, size).dplyr::mutate(). Be brave and deal with it.(sampled_iris <- nested_iris %>%
mutate(samp = map2(data, n, sample_n)))
#> # A tibble: 3 x 4
#> Species data n samp
#> <fct> <list<df[,4]>> <dbl> <list>
#> 1 setosa [50 × 4] 2 <tibble [2 × 4]>
#> 2 versicolor [50 × 4] 5 <tibble [5 × 4]>
#> 3 virginica [50 × 4] 3 <tibble [3 × 4]>What came back? More Species-specific data frames.
We are in that uncomfortable intermediate state, with two list-columns: the original data and the sampled data, samp. Let’s get back to a normal data frame!
Species and samp variables.samp and replicates Species as necessary.sampled_iris %>%
select(-data) %>%
unnest(samp)
#> # A tibble: 10 x 6
#> Species n Sepal.Length Sepal.Width Petal.Length Petal.Width
#> <fct> <dbl> <dbl> <dbl> <dbl> <dbl>
#> 1 setosa 2 4.6 3.2 1.4 0.2
#> 2 setosa 2 5.1 3.8 1.5 0.3
#> 3 versicolor 5 6 3.4 4.5 1.6
#> 4 versicolor 5 6 2.2 4 1
#> 5 versicolor 5 5.7 2.8 4.5 1.3
#> 6 versicolor 5 6.9 3.1 4.9 1.5
#> 7 versicolor 5 7 3.2 4.7 1.4
#> 8 virginica 3 5.8 2.7 5.1 1.9
#> 9 virginica 3 5.8 2.7 5.1 1.9
#> 10 virginica 3 6.3 2.9 5.6 1.8Again, from the top, with no exposition:
iris %>%
group_by(Species) %>%
nest() %>%
ungroup() %>%
mutate(n = c(2, 5, 3)) %>%
mutate(samp = map2(data, n, sample_n)) %>%
select(-data) %>%
unnest(samp)
#> # A tibble: 10 x 6
#> Species n Sepal.Length Sepal.Width Petal.Length Petal.Width
#> <fct> <dbl> <dbl> <dbl> <dbl> <dbl>
#> 1 setosa 2 4.7 3.2 1.6 0.2
#> 2 setosa 2 4.8 3.1 1.6 0.2
#> 3 versicolor 5 6.7 3.1 4.7 1.5
#> 4 versicolor 5 5.1 2.5 3 1.1
#> 5 versicolor 5 5.2 2.7 3.9 1.4
#> 6 versicolor 5 5.7 3 4.2 1.2
#> 7 versicolor 5 5.6 2.7 4.2 1.3
#> 8 virginica 3 7.6 3 6.6 2.1
#> 9 virginica 3 5.6 2.8 4.9 2
#> 10 virginica 3 6.4 3.1 5.5 1.8A base R solution, with some marginal comments:
split_iris <- split(iris, iris$Species) # why can't Species be found in iris?
# where else would it be found?
str(split_iris) # split_iris ~= nested_iris[["data"]]
#> List of 3
#> $ setosa :'data.frame': 50 obs. of 5 variables:
#> ..$ Sepal.Length: num [1:50] 5.1 4.9 4.7 4.6 5 5.4 4.6 5 4.4 4.9 ...
#> ..$ Sepal.Width : num [1:50] 3.5 3 3.2 3.1 3.6 3.9 3.4 3.4 2.9 3.1 ...
#> ..$ Petal.Length: num [1:50] 1.4 1.4 1.3 1.5 1.4 1.7 1.4 1.5 1.4 1.5 ...
#> ..$ Petal.Width : num [1:50] 0.2 0.2 0.2 0.2 0.2 0.4 0.3 0.2 0.2 0.1 ...
#> ..$ Species : Factor w/ 3 levels "setosa","versicolor",..: 1 1 1 1 1 1 1 1 1 1 ...
#> $ versicolor:'data.frame': 50 obs. of 5 variables:
#> ..$ Sepal.Length: num [1:50] 7 6.4 6.9 5.5 6.5 5.7 6.3 4.9 6.6 5.2 ...
#> ..$ Sepal.Width : num [1:50] 3.2 3.2 3.1 2.3 2.8 2.8 3.3 2.4 2.9 2.7 ...
#> ..$ Petal.Length: num [1:50] 4.7 4.5 4.9 4 4.6 4.5 4.7 3.3 4.6 3.9 ...
#> ..$ Petal.Width : num [1:50] 1.4 1.5 1.5 1.3 1.5 1.3 1.6 1 1.3 1.4 ...
#> ..$ Species : Factor w/ 3 levels "setosa","versicolor",..: 2 2 2 2 2 2 2 2 2 2 ...
#> $ virginica :'data.frame': 50 obs. of 5 variables:
#> ..$ Sepal.Length: num [1:50] 6.3 5.8 7.1 6.3 6.5 7.6 4.9 7.3 6.7 7.2 ...
#> ..$ Sepal.Width : num [1:50] 3.3 2.7 3 2.9 3 3 2.5 2.9 2.5 3.6 ...
#> ..$ Petal.Length: num [1:50] 6 5.1 5.9 5.6 5.8 6.6 4.5 6.3 5.8 6.1 ...
#> ..$ Petal.Width : num [1:50] 2.5 1.9 2.1 1.8 2.2 2.1 1.7 1.8 1.8 2.5 ...
#> ..$ Species : Factor w/ 3 levels "setosa","versicolor",..: 3 3 3 3 3 3 3 3 3 3 ...
(n <- c(2, 5, 3)) # Species data and n are only 'in sync'
#> [1] 2 5 3
# due to my discipline / care
# not locked safely into a data frame
(group_sizes <- vapply(split_iris, nrow, integer(1))) # also floating free
#> setosa versicolor virginica
#> 50 50 50
(sampled_obs <- mapply(sample, group_sizes, n)) # I'm floating free too!
#> $setosa
#> [1] 36 14
#>
#> $versicolor
#> [1] 28 43 22 24 13
#>
#> $virginica
#> [1] 48 3 25
get_rows <- function(df, rows) df[rows, , drop = FALSE] # custom function
# drop = FALSE required to avoid
# nasty surprise in case of n = 1
(sampled_iris <- # god help you if forget SIMPLIFY = FALSE
mapply(get_rows, split_iris, sampled_obs, SIMPLIFY = FALSE))
#> $setosa
#> Sepal.Length Sepal.Width Petal.Length Petal.Width Species
#> 36 5.0 3.2 1.2 0.2 setosa
#> 14 4.3 3.0 1.1 0.1 setosa
#>
#> $versicolor
#> Sepal.Length Sepal.Width Petal.Length Petal.Width Species
#> 78 6.7 3.0 5.0 1.7 versicolor
#> 93 5.8 2.6 4.0 1.2 versicolor
#> 72 6.1 2.8 4.0 1.3 versicolor
#> 74 6.1 2.8 4.7 1.2 versicolor
#> 63 6.0 2.2 4.0 1.0 versicolor
#>
#> $virginica
#> Sepal.Length Sepal.Width Petal.Length Petal.Width Species
#> 148 6.5 3.0 5.2 2.0 virginica
#> 103 7.1 3.0 5.9 2.1 virginica
#> 125 6.7 3.3 5.7 2.1 virginica
do.call(rbind, sampled_iris) # :( do.call()
#> Sepal.Length Sepal.Width Petal.Length Petal.Width Species
#> setosa.36 5.0 3.2 1.2 0.2 setosa
#> setosa.14 4.3 3.0 1.1 0.1 setosa
#> versicolor.78 6.7 3.0 5.0 1.7 versicolor
#> versicolor.93 5.8 2.6 4.0 1.2 versicolor
#> versicolor.72 6.1 2.8 4.0 1.3 versicolor
#> versicolor.74 6.1 2.8 4.7 1.2 versicolor
#> versicolor.63 6.0 2.2 4.0 1.0 versicolor
#> virginica.148 6.5 3.0 5.2 2.0 virginica
#> virginica.103 7.1 3.0 5.9 2.1 virginica
#> virginica.125 6.7 3.3 5.7 2.1 virginicaIMO the base R solution requires much greater facility with R programming and data structures to get it right. It feels more like programming than data analysis.