Replacing colum with values from lookup table in R

Question

I need to replace the values of a column model in ddata:

> unique(ddata$model)
[1] "GT-I9001"    "iPhone5,2"    "iPhone3,1"                        
[4] "iPhone4,1"   "GT-I9300"     "Nexus 4"                         
[7] "iPhone2,1"   "VS840 4G"     "HTC One X+"  
...

with those from a lookup table, which I imported as

> devices
...
15 iPhone1,1         iPhone 2G,
16 iPhone1,2         iPhone 3G,
17 iPhone2,1         iPhone 3GS,
18 iPhone3,1         iPhone 4,
19 iPhone3,2         iPhone 4,
20 iPhone3,3         iPhone 4,
21 iPhone4,1         iPhone 4S,
22 iPhone5,1         iPhone 5,
23 iPhone5,2         iPhone 5,
...

So that, for example, iPhone4,1 becomes iPhone4S. If a key is not present in the lookup table, I want to keep the original value of ddata$model, so for example the original Nexus 4 remains Nexus 4.

I tried with merge, but I don't know how to address the case where a key is missing:

ddata2 = merge(ddata,devices,by="model", all.x=T)

The problem being that ddata2 has N/A for those values that don't have an entry in devices.

Same question but with a poorer title: http://stackoverflow.com/questions/17324851/replace-multible-data-codes-in-a-datafram-with-names/17325215#17325215 — Tyler Rinker, Jul 01 '13 at 19:56

Tyler Rinker · Answer 1 · 2014-02-27T01:08:22.780

Here's one approach using qdap:

## key <- readLines(n=9)
## iPhone1,1         iPhone 2G,
## iPhone1,2         iPhone 3G,
## iPhone2,1         iPhone 3GS,
## iPhone3,1         iPhone 4,
## iPhone3,2         iPhone 4,
## iPhone3,3         iPhone 4,
## iPhone4,1         iPhone 4S,
## iPhone5,1         iPhone 5,
## iPhone5,2         iPhone 5,
## 
## library(qdap)
## key <- colSplit(key, "         ")
## 
## ddata <- data.frame(model = c("GT-I9001",    "iPhone5,2",    "iPhone3,1",                        
##     "iPhone4,1",   "GT-I9300",     "Nexus 4",                         
##     "iPhone2,1",   "VS840 4G",     "HTC One X+"))

library(qdap)

ddata$model <- as.character(ddata$model)

ddata$model <- lookup(ddata$model, key, missing =NULL) ## OR
ddata$model <- ddata$model %l+% key

Please note that your data was difficult to read in. Please use dput for odd data sets. Also I believe a similar question was asked recently but can't find it.

score 1 · Accepted Answer · answered Jul 01 '13 at 20:55

1

I would use merge like above, then follow with:

ddata2 <- within( ddata2, model[ is.na(model2) ] <- model[ is.na(model2) ] )

Assuming that the 2nd column in devices is called model2.

answered Jul 01 '13 at 20:55

Greg Snow

48,497
6
83
110

I'm going to accept this one as it doesn't require the extra `qdap` package. – gozzilli Jul 01 '13 at 21:15

Replacing colum with values from lookup table in R

2 Answers2