CODEWITHSUNDEEP
r
Rilcon42
Rilcon42

Reputation: 9765

recoding numbers to strings does not replace all values

Can someone explain why only some of the numbers are being converted to the strings high and low?

set.seed(1)
df <- data.frame(npitotal=ceiling(runif(30,0,40)))
df$npitotal[df$npitotal >= 22] <- "high"
df$npitotal[df$npitotal < 22] <- "low"

df$npitotal
# [1] "low"  "low"  "high" "high" "9"    "high" "high" "high"
# [9] "high" "3"    "9"    "8"    "high" "low"  "high" "low" 
#[17] "high" "high" "low"  "high" "high" "9"    "high" "6"   
#[25] "low"  "low"  "low"  "low"  "high" "low"

Upvotes: 2

Views: 45

Answers (1)

Rilcon42
Rilcon42

Reputation: 9765

My eventual solution was to use recode in the car package:

library(car)
recode(df$npitotal,"0:21='low';22:40='high'")

Note that 0:21 is [0,21] (so 20.99 evaluates to "low: in this case)


Thanks to @Jay for explaining the reason this causes a problem:

After you do your first conversion it becomes a character vector - so the second less than is not being evaluated numerically but as a character.

Upvotes: 2

Related Questions