How can I take log of particular columns? - r

Question

I have this data
structure(list(DATE = 19620101:19620106, PRECIP = c(10.54, 6.39,
0.01, 0, 0.02, 20.94), OBS_Q = c(2.39, 2.38, 2.22, 2.24, 2.26,
5.13), swb = c(4.11, 3.92, 3.8, 3.8, 3.77, 7.16), gr4j = c(3.7,
4.24, 3.73, 3.24, 2.98, 4.93), isba = c(4.82, 3.44, 4.18, 3.01,
2.88, 6.35), noah = c(3.11, 2.84, 2.57, 2.59, 2.84, 4.83), sac = c(2.83,
2.84, 2.73, 2.87, 2.94, 6), swap = c(2.16, 2.56, 2.08, 2.19,
3.65, 4.43), vic.mm.day. = c(3.44, 3.14, 3.37, 3.15, 2.88, 4.52
)), .Names = c("DATE", "PRECIP", "OBS_Q", "swb", "gr4j", "isba",
"noah", "sac", "swap", "vic.mm.day."), row.names = c(NA, 6L), class = "data.frame")
And a function that does three things
Subsets based on what 'model' is mentioned(one of the column names of df)
Outputs dfcal with 3 columns
Outputs dfcallog with the log of those values
The function
dataprep.allcal<-function(df,model){
dfcal<<-subset(df, select=c("DATE",model, "OBS_Q"))
dfcallog<-subset(df, select=c("DATE",model, "OBS_Q"))
cols<-colnames(dfcallog)
dfcallog[cols] <<- log(dfcallog[cols])
}
dataprep.allcal(df=df,model='sac')
The problem with this, it converts the date to log as well.
Can you tell me how I can apply the log only on model and OBS_Q column?

Answer 1

This is a way to do it by excluding DATE from the cols variable:
dataprep.allcal<-function(df,model){
dfcal<<-subset(df, select=c("DATE",model, "OBS_Q"))
#need to use the <<- operator below
dfcallog<<-subset(df, select=c("DATE",model, "OBS_Q"))
cols<-colnames(dfcallog)
cols<-cols[!cols %in% 'DATE']
dfcallog[cols] <<- log(dfcallog[cols])
}
dataprep.allcal(df=df,model='sac')
Output:
> dfcal
DATE sac OBS_Q
1 19620101 2.83 2.39
2 19620102 2.84 2.38
3 19620103 2.73 2.22
4 19620104 2.87 2.24
5 19620105 2.94 2.26
6 19620106 6.00 5.13
> dfcallog
DATE sac OBS_Q
1 19620101 1.040277 0.8712934
2 19620102 1.043804 0.8671005
3 19620103 1.004302 0.7975072
4 19620104 1.054312 0.8064759
5 19620105 1.078410 0.8153648
6 19620106 1.791759 1.6351057
Note:
I don't know if this is a typo but you should have dfcallog<-subset(df, select=c("DATE",model, "OBS_Q")) with a <<- operator because using a <- operator at the beginning and then a <<- operator at the last command in the function will produce an error because <<- looks at the parent.environment to find a variable and thus you would get an object not found error.