3

I'm using aregImpute to impute missing values on a R dataframe (bn_df).

The code is this:

library(Hmisc)
impute_arg <- aregImpute(~ TI_Perc + AS_Perc + 
                         CD_Perc + CA_Perc + FP_Perc, 
                         data = bn_df, n.impute = 5)

It works fine.

The problem is after. In putting the values back into the original dataframe.

I can do it, just not in a very elegant way. I basically have to copy/paste the following line for all columns:

bn_df$CD_Perc[impute_arg$na$CD_Perc] <- impute_arg$imputed$CD_Perc[,1]
bn_df$FP_Perc[impute_arg$na$FP_Perc] <- impute_arg$imputed$FP_Perc[,1]
...

This works. But there has to be a more efficient way to accomplish this without copy/paste for all columns.

Any ideas?

makeyourownmaker
  • 1,558
  • 2
  • 13
  • 33
BrunoPT
  • 37
  • 1
  • 5

1 Answers1

8

You can use function impute.transcan. Since you have not provided the data, I have copied an example from aregImpute's documentation.

# The data
x1 <- factor(sample(c('a','b','c'),1000,TRUE))
x2 <- (x1=='b') + 3*(x1=='c') + rnorm(1000,0,2)
x3 <- rnorm(1000)
y  <- x2 + 1*(x1=='c') + .2*x3 + rnorm(1000,0,2)
orig.x1 <- x1[1:250]
orig.x2 <- x2[251:350]

# Insert NAs
x1[1:250] <- NA
x2[251:350] <- NA

# Create a data frame 
d <- data.frame(x1,x2,x3,y)
# Find value of nk that yields best validating imputation models
# tlinear=FALSE means to not force the target variable to be linear

# Use imputation 
f <- aregImpute(~y + x1 + x2 + x3, nk=c(0,3:5), tlinear=FALSE,
                data=d, B=10) # normally B=75

# Get the imputed values
imputed <-impute.transcan(f, data=d, imputation=1, list.out=TRUE, pr=FALSE, check=FALSE)

# convert the list to the database
imputed.data <- as.data.frame(do.call(cbind,imputed))

# arrange the columns accordingly
imputed.data <- imputed.data[, colnames(d), drop = FALSE]
discipulus
  • 2,665
  • 3
  • 34
  • 51
  • Thank you! It worked perfectly. Indeed I probably skipped that function. – BrunoPT Feb 03 '17 at 09:24
  • 6
    Is there any way we can replace value with average of multiple imputations rather than any specific number (imputation=1)? – Sourabh Apr 13 '17 at 07:27