fifer (version 1.0)

stratified: Sample from a data.frame according to a stratification variable

Description

The stratified function samples from a data.frame in which one of the columns can be used as a "stratification" or "grouping" variable. The result is a new data.frame with the specified number of samples from each group.

Usage

stratified(df, group, size, select = NULL, seed = NULL, ...)

Arguments

df

The source data.frame.

group

Your grouping variables. Generally, if you are using more than one variable to create your "strata", you should list them in the order of slowest varying to quickest varying. This can be a vector of names or column indexes.

size

The desired sample size.

• If size is a value between 0 and 1 expressed as a decimal, size is set to be proportional to the number of observations per group.
• If size is a single positive integer, it will be assumed that you want the same number of samples from each group.
• If size is a vector, the function will check to see whether the length of the vector matches the number of groups and use those specified values as the desired sample sizes. The values in the vector should be in the same order as you would get if you tabulated the grouping variable (usually alphabetic order); alternatively, you can name each value to ensure it is properly matched.

select

A named list containing levels from the "group" variables in which you are interested. The list names must be present as variable names for the input data.frame.

seed

The seed that you want to use (using set.seed within the function, if any. Defaults to NULL.

...

Further arguments to be passed to the sample function.

References

The evolution of this function can be traced at the following links. The version in this package is entirely reworked and does not require an additional package to be loaded.

• http://news.mrdwab.com/2011/05/15/stratified-random-sampling-in-r-beta/
• http://news.mrdwab.com/2011/05/20/stratified-random-sampling-in-r-from-a-data-frame/
• http://stackoverflow.com/a/9714207/1270695

Examples

# Generate a couple of sample data.frames to play with
set.seed(1)
dat1 <- data.frame(ID = 1:100,
              A = sample(c("AA", "BB", "CC", "DD", "EE"), 100, replace = TRUE),
              B = rnorm(100), C = abs(round(rnorm(100), digits=1)),
              D = sample(c("CA", "NY", "TX"), 100, replace = TRUE),
              E = sample(c("M", "F"), 100, replace = TRUE))
dat2 <- data.frame(ID = 1:20,
              A = c(rep("AA", 5), rep("BB", 10),
                    rep("CC", 3), rep("DD", 2)))
# What do the data look like in general?
summary(dat1)
summary(dat2)

# Let's take a 10% sample from all -A- groups in dat1, seed = 1
stratified(dat1, "A", .1, seed = 1)

# Let's take a 10% sample from only "AA" and "BB" groups from -A- in dat1, seed = 1
stratified(dat1, "A", .1, select = list(A = c("AA", "BB")), seed = 1)

# Let's take 5 samples from all -D- groups in dat1,
#   seed = 1, specified by column number
stratified(dat1, group = 5, size = 5, seed = 1)

# Let's take a sample from all -A- groups in dat1, seed = 1,
#   where we specify the number wanted from each group
stratified(dat1, "A", size = c(3, 5, 4, 5, 2), seed = 1)

# Use a two-column strata: -E- and -D-
#   -E- varies more slowly, so it is better to put that first
stratified(dat1, c("E", "D"), size = .15, seed = 1)

# Use a two-column strata (-E- and -D-) but only interested in
#   cases where -E- == "M"
stratified(dat1, c("E", "D"), .15, select = list(E = "M"), seed = 1)

## As above, but where -E- == "M" and -D- == "CA" or "TX"
stratified(dat1, c("E", "D"), .15,
     select = list(E = "M", D = c("CA", "TX")), seed = 1)

# Use a three-column strata: -E-, -D-, and -A-
s.out <- stratified(dat1, c("E", "D", "A"), size = 2, seed = 1)

list(head(s.out), tail(s.out))

# How many samples were taken from each strata?
table(interaction(s.out[c("E", "D", "A")]))

# Can we verify the message about group sizes?
names(which(table(interaction(dat1[c("E", "D", "A")])) < 2))

names(which(table(interaction(s.out[c("E", "D", "A")])) < 2))