CODEWITHSUNDEEP
r
gojomoso
gojomoso

Reputation: 163

R Evaluate Multiple Columns by Proportion

df1:

a = c(2, 3, 5, 8, 10, 12) 
b = c("NA", "bb", "cc", "aa", "bb", "aa") 
c = c("bb", "aa", "bb", "cc", "aa", "aa")
d = c("aa", "cc", "bb", "aa", "aa", "aa")
e = c(TRUE, FALSE, TRUE, FALSE, TRUE, FALSE) 
df1 = data.frame(a, b, c, d, e)

Looking to evaluate the proportion of all values combined in b, c, d and then change any category with a proportion below 20% to "Rare".

Output:

a  b    c    d    e
2  NA   bb   aa   true
3  bb   aa   rare false
5  rare bb   bb   true
8  aa   rare aa   false
10 bb   aa   aa   true
12 aa   aa   aa   false

Upvotes: 1

Views: 37

Answers (1)

Gregor Thomas
Gregor Thomas

Reputation: 145985

Here's a base R approach:

# convert string "NA" to actual missing values NA
df1[df1 == "NA"] = NA

cols = c("b", "c", "d")

freq = prop.table(table(unlist(df1[cols])))
make_rare = names(freq)[freq < 0.2]

df1[cols] = lapply(df1[cols], function(x) replace(x, x %in% make_rare, "rare"))

df1
#    a    b    c    d     e
# 1  2 <NA>   bb   aa  TRUE
# 2  3   bb   aa rare FALSE
# 3  5 rare   bb   bb  TRUE
# 4  8   aa rare   aa FALSE
# 5 10   bb   aa   aa  TRUE
# 6 12   aa   aa   aa FALSE

Upvotes: 1

Related Questions