Subsetting data based on variable start date in R - r

Question

I am trying to write a bit of code that will grab a portion of a data frame based on start date, where each start date is different for each user.
Suppose I have the following data.frames (in reality my dataset is several orders of magnitude larger, but this will suffice as a sample set)
df1:
> df
name start.date
1 Allison 2013-03-16
2 Andrew 2013-03-16
3 Carl 2013-03-16
4 Dora 2013-03-17
5 Hilary 2013-03-17
6 Louis 2013-03-18
7 Mary 2013-03-19
8 Mickey 2013-03-20
And df2:
> df2
names X03.16.2013 X03.17.2013 X03.18.2013 X03.19.2013
2001 Allison 5 5 0 0
2002 Andrew 2 0 0 0
2003 Carl 10 8 11 10
2004 Dora 0 4 0 0
2005 Hilary 0 3 5 0
2006 Louis 0 0 8 3
2007 Mary 0 0 0 7
2008 Mickey 0 0 0 0
I merged these two data frames into one called tmp:
>tmp
name start.date X03.16.2013 X03.17.2013 X03.18.2013 X03.19.2013
1 Allison 2013-03-16 5 5 0 0
2 Andrew 2013-03-16 2 0 0 0
3 Carl 2013-03-16 10 8 11 10
4 Dora 2013-03-17 0 4 0 0
5 Hilary 2013-03-17 0 3 5 0
6 Louis 2013-03-18 0 0 8 3
7 Mary 2013-03-19 0 0 0 7
8 Mickey 2013-03-20 0 0 0 0
I also have a list of the column names of df2 converted to dates:
>dts
[1] "2014-03-16" "2014-03-17" "2014-03-18" "2014-03-19"
I thought one way to approach this problem is to change all the zero entries in df2 that occur before each user's start date to NA using the following nested loops:
for (i in 1:dim(tmp)[1]){
for (j in 1:length(dts)){
for (z in 4:dim(tmp)[2]){
if (dts[j]< tmp$Date.of.Sign.Up[i]){
tmp[i,z]<-NA
} else {tmp[i,z]<-tmp[i,z]}
}
}
}
The trouble with this loop is that 1. It will run infinitely and 2. Doesn't work. It is changing all the values in tmp from tmp[,3:end] to zero, regardless of start date. Ideally I would end up with something like this:
name start.date X03.16.2013 X03.17.2013 X03.18.2013 X03.19.2013
Allison 2013-03-16 5 5 0 0
Andrew 2013-03-16 2 0 0 0
Carl 2013-03-16 10 8 11 10
Dora 2013-03-17 NA 4 0 0
Hilary 2013-03-17 NA 3 5 0
Louis 2013-03-18 NA NA 8 3
Mary 2013-03-19 NA NA NA 7
Mickey 2013-03-20 NA NA NA NA
Any suggestions? Thank you in advance!

Answer 1

You may reshape 'tmp' to long format, convert the former headers to dates, compare them with the start dates, and insert NA when start dates occur after the 'header dates':
library(reshape2)
# melt data from wide to long format
df3 <- melt(tmp, id.vars = c("name", "start.date"))
# convert 'variable' to class Date
df3$variable <- as.Date(df3$variable, format = "X%m.%d.%Y")
# compare start dates with 'variable dates' and insert NA
df3$value[df3$start.date > df3$variable] <- NA
# reshape back to wide
dcast(df3, name + start.date ~ variable)
# name start.date 2013-03-16 2013-03-17 2013-03-18 2013-03-19
# 1 Allison 2013-03-16 5 5 0 0
# 2 Andrew 2013-03-16 2 0 0 0
# 3 Carl 2013-03-16 10 8 11 10
# 4 Dora 2013-03-17 NA 4 0 0
# 5 Hilary 2013-03-17 NA 3 5 0
# 6 Louis 2013-03-18 NA NA 8 3
# 7 Mary 2013-03-19 NA NA NA 7
# 8 Mickey 2013-03-20 NA NA NA NA
Another possibility where we loop over the 'date column names' in 'tmp':
dates <- names(tmp)[-c(1, 2)]
tmp[ , -c(1, 2)] <- sapply(dates, function(x){
date <- as.Date(x, format = "X%m.%d.%Y")
tmp[ , x][df2$start.date > date] <- NA
tmp[ , x]
})
tmp
# name start.date X03.16.2013 X03.17.2013 X03.18.2013 X03.19.2013
# 1 Allison 2013-03-16 5 5 0 0
# 2 Andrew 2013-03-16 2 0 0 0
# 3 Carl 2013-03-16 10 8 11 10
# 4 Dora 2013-03-17 NA 4 0 0
# 5 Hilary 2013-03-17 NA 3 5 0
# 6 Louis 2013-03-18 NA NA 8 3
# 7 Mary 2013-03-19 NA NA NA 7
# 8 Mickey 2013-03-20 NA NA NA NA