0

I have been working on this project and I am stuck in the following:

I have 7 columns on which over 30% of the rows are NAs. All my columns are numeric, by the way.

On these High Missing Values Columns I want to create 4 new columns base on the values of the these columns' quantiles.

1st column- input 1 in rows which contains data; 0 otherwise. 2nd column- input 1 in rows below the first quantile; 0 otherwise. 3rd column- input 1 in rows that are in the 2nd quantile range; 0 otherwise. 4th column- input 1 in rows that are above the 3rd quantile; 0 otherwise.

I got the first column. But the rest, based on the quantiles' threshold value has been a challenge. Here is what I have so far...

My next 3 columns are base on just 3 quantiles: 33.33333%, 66.66667% and 100%

quantile(High_NAS_set1$EFX, prob=c(33/99,66/99,99/99),na.rm=TRUE)



#1st column: assign 1 for a row that contains data; 0 otherwise

New.EFX_<-High_NAS_set1$EFX #creating a new column
New.EFX[!is.na(New.EFX)]<-1
New.EFX[is.na(New.EFX)]<-0

#2nd Column:assign 1 in rows below the first quantile; 0 otherwise

New.EFX2_<-High_NAS_set1$EFX #creating a new column
quant<-quantile(New2.EFX_Emp,probs=33/99,na.rm=TRUE)
which(New2.EFX_Emp_Total<=quant)<-1  # assign 1 for rows which indexes are below quant
which(New2.EFX_Emp_Total!=quant)<-0

The last 2 lines are giving me an error:

Error in which(New2.EFX_Emp_Total <= quant) <- 1 : 
  could not find function "which<-"

Any help will be really appreciated. Thanks, Jean

user2913919
  • 21
  • 2
  • Including sample input and desired output would make a more complete [reproducible example](http://stackoverflow.com/questions/5963269/how-to-make-a-great-r-reproducible-example) which would make it easier to help you. – MrFlick Jul 06 '14 at 00:32
  • `New.EFX_<-High_NAS_set1$EFX` would NOT create a new column. If you did have column named 'New.EFX' then you would need to write: `High_NAS_set1$New.EFX[ !is.na(High_NAS_set1$EFX) ] <- 1`. I'm afraid the rest of your request regarding "quantile thresholds" made no sense to me, and I suspect that was why @MrFlick asked for the example. – IRTFM Jul 06 '14 at 17:39

0 Answers0