CODEWITHSUNDEEP
r
Rene Chan
Rene Chan

Reputation: 985

Apply function with three attributes in R

I have a table with three columns ABC, EFG, HIJ. I would like to create a fourth column KLM which is a function of the conditional value of ABC, and a operation result on EFG and HIJ.

For now I am using a loop that takes about 15 minutes over 400,000 rows. And that does not seem very R to me. There must be a way to do this significantly less time:

for (i in 1:nrow(df)){
  if(is.na(df$ABC[i]) == FALSE ){
    df$KLM[i] <- as.numeric(df$EFG[i] * df$HIJ[i])
  } else {
    df$KLM[i] = NaN
  }
}

I have added the df: 

ABC = c("NaN", 232,234,233,232.5)
EFG = c(12,12,12,12,12)
HIJ = c(10.75, 10.95, 11.25, 10.85, 10.55)
KLM = c(0,0,0,0,0)

df <- as.data.frame(cbind(ABC, EFG, HIJ, KLM))
df < unfactor(df)


> df
    ABC EFG   HIJ KLM
1   NaN  12 10.75   0
2   232  12 10.95   0
3   234  12 11.25   0
4   233  12 10.85   0
5 232.5  12 10.55   0

Does anyone knows how to simplify and make more efficient please ?

Upvotes: 1

Views: 38

Answers (1)

tushaR
tushaR

Reputation: 3116

@jogo's solution mentioned in the comments is the best vectorized solution for data.frame.

Using data.table it can be optimized as follows:

dt = as.data.table(df)
dt[,`:=`(KLM=NaN)]
set(x = dt, i =which(!is.na(dt$ABC)),j="KLM",value = as.numeric(EFG * HIJ))

Upvotes: 1

Related Questions