Delete repeated TIME points in a dataframe - r

Question

I have a simple goal that I want to achieve in my data frame that looks like this:
ID TIME AMT
1 0 100
1 1 0
1 2 0
1 2 50
1 3 0
2 0 50
2 1 0
2 2 0
2 2 100
2 3 0
How do I subset the df for unique TIME (i.e. get rid of the repeated time point that has AMT=0? To make it clearer: I want to remove duplicate TIME rows that has AMT=0.

Answer 1

It is not entirely clear what you're asking. I think what you want is, for each unique ID value, eliminate duplicate TIME rows, and if a duplicate row has AMT=0, prefer to delete that row rather than another duplicate (with the same TIME value) that has AMT!=0.
The best way to do this is actually to call aggregate(), and group by both ID and TIME, taking the max() of all the AMT values in all the duplicates in a group (thus this will work for duplicate groups that have more than two rows, if such existed):
df <- data.frame(id=c(1,1,1,1,1,2,2,2,2,2), time=c(0,1,2,2,3,0,1,2,2,3), amt=c(100,0,0,50,0,50,0,0,100,0) );
df;
## id time amt
## 1 1 0 100
## 2 1 1 0
## 3 1 2 0
## 4 1 2 50
## 5 1 3 0
## 6 2 0 50
## 7 2 1 0
## 8 2 2 0
## 9 2 2 100
## 10 2 3 0
aggregate(amt~id+time, df, max );
## id time amt
## 1 1 0 100
## 2 2 0 50
## 3 1 1 0
## 4 2 1 0
## 5 1 2 50
## 6 2 2 100
## 7 1 3 0
## 8 2 3 0
As you can see, the order got a little messed up, but you could easily fix that with a call to order() afterward:
df2 <- aggregate(amt~id+time, df, max );
df2[order(df2$id,df2$time),];
## id time amt
## 1 1 0 100
## 3 1 1 0
## 5 1 2 50
## 7 1 3 0
## 2 2 0 50
## 4 2 1 0
## 6 2 2 100
## 8 2 3 0

Answer 2

It is not entirely clear from the description, how we want to remove the duplicated elements. Suppose if there are duplicates for 'TIME', 'ID', but the 'AMT' element is neither zero nor maximum value. If we need to remove only the '0' values per combination,
library(data.table)
res1 <- setDT(df1)[, if(all(AMT==0)) .SD[1L] else .SD[AMT!=0], list(TIME,ID)]
res1[order(TIME)]
# TIME ID AMT
#1: 0 1 100
#2: 0 2 50
#3: 1 1 0
#4: 1 2 0
#5: 2 1 50
#6: 2 2 100
#7: 3 1 0
#8: 3 2 0
or if the idea of removing the duplicates was as assumed by #bgoldst, an equivalent option using data.table is
res2 <- setDT(df1)[, list(amt=max(AMT)), list(TIME, ID)]
res2[order(TIME)]
# TIME ID amt
#1: 0 1 100
#2: 0 2 50
#3: 1 1 0
#4: 1 2 0
#5: 2 1 50
#6: 2 2 100
#7: 3 1 0
#8: 3 2 0
data
df1 <- structure(list(ID = c(1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L),
TIME = c(0L, 1L, 2L, 2L, 3L, 0L, 1L, 2L, 2L, 3L), AMT = c(100L,
0L, 0L, 50L, 0L, 50L, 0L, 0L, 100L, 0L)), .Names = c("ID",
"TIME", "AMT"), class = "data.frame", row.names = c(NA, -10L))