balancer: Function to balance a sample with regards to a dominant...

In talentlens/talentlens: R package for team TalentLens

Description

Takes a group variable (factor vector) f and the desired maximum proportion of the largest subgroup (factor level) p, samples randomly from the largest group and returns a logical vector, indicating which cases to retain to reach the desired proportion of the largest group.

Usage

  balancer(f, p, silent = TRUE)

Arguments

f	Factor vector representing a group variable to be balanced, where the levels represent groups/categories in the data. For example, this could be a column in a dataframe, indicating the gender of the participants.
p	Desired maximum proportion for the largest subgroup. For example, if we want a balanced sample with regards to gender, we may want to allow for no more than 50 % of the total sample to be from either gender.
silent	If TRUE (default), messages are not printed to the console. Set to FALSE if you want to see how many cases are dropped randomly.

Details

Formula for determining the necessary reduction of the sample size (delta_n) for the dominant subgroup:

delta_n = N (p2 - p1) / (1 - p2)

Value

Logical vector of the same length as f, with values TRUE for cases to keep and FALSE for cases to drop.

Author(s)

Morgan Strom

Examples

##Generate factor vector with unbalanced levels
f <- factor(x = c(rep("A", 20), rep("B", 50), rep("C", 30)))
table(f)

#Calculate proportions of the groups
table(f) / length(f)

#Use balancer() function to reduce the largest group to 40%

#First, create the index vector
i <- balancer(f, p = 0.40, silent = FALSE)

#Second, subset the factor vector and store in a new object, f2
f2 <- f[i]

#Calculate new proportions
table(f2) / length(f2)

talentlens/talentlens documentation built on May 31, 2019, 2:52 a.m.