用R中的原始行替换特定行的汇总结果

Question

I am using aggregate function to aggregate results for a subset of my data set. I want the final results to be replaces with the original rows (reference rows for aggregate). How can I do that? Here is a sample data:

 Day  hour    Case   Time
 Sat  7       2    35
 Sun  8       8    125
 Sun  9       10   145
 Mon  10      15   18
 Mon  11      17   167
 Mon  12      20   220
 Mon  13      25   135
 Mon  14      14   167

I used the following line of code to aggregate Case and Time Values for "Sat" and "Sun"

aggregate(cbind(Case,Time)~Day,data=subset(TestData,Day == 'Sat' |Day == 'Sun' ),sum)

which works prefectly correct. However, I wonder how I can replace rows 2,3 and four of my sample data with the aggregate result I get. I want the final result to be like this:

       Day  hour    Case   Time
 Sat  7       2    35
 Sun  8       18   270
 Mon  10      15   18
 Mon  11      17   167
 Mon  12      20   220
 Mon  13      25   135
 Mon  14      14   167  

Thanks

Answer 1

We can use data.table to do this. We select the columns that we need to get the sum value ('nm1'). Convert the 'data.frame' to 'data.table' ( setDT(df1) ), specify the 'i' part with the logical condition to exclude other rows ( Day %in% c('Sat', 'Sun') , using .SDcols we select the columns for sum , loop ( lapply ) through the Subset of Data.Table ( .SD ), and assign ( := ) the output to the columns in 'nm1' and the rows specified in the 'i'.

library(data.table)
nm1 <- c('Case', 'Time')
setDT(df1)[Day %in% c('Sat', 'Sun'), (nm1) := lapply(.SD, sum),
                        Day, .SDcols=nm1]

If we need only the unique rows, we can use the unique from data.table with the by option

unique(df1, by=c('Case', 'Time'))
#   Day hour Case Time
#1: Sat    7    2   35
#2: Sun    8   18  270
#3: Mon   10   15   18
#4: Mon   11   17  167
#5: Mon   12   20  220
#6: Mon   13   25  135
#7: Mon   14   14  167

---

Or if we are using the OP's aggregate code, we can merge the 'r1' with the original dataset ('df1'), replace the 'rows' with the logical index derived from 'NA' values after the merge , subset the columns, remove the duplicated rows and get the output

r1 <- aggregate(cbind(Case,Time)~Day,data=subset(df1,
                Day == 'Sat' |Day == 'Sun' ),sum)

r2 <- merge(df1, r1, by='Day', all.x=TRUE)
r2[indx, c('Case.x', 'Time.x')] <- r2[indx, c('Case.y', 'Time.y')]

We select only the columns that are need

r3 <- r2[1:4]

Remove the duplicate rows in the 'Case', 'Time' column

r3[!duplicated(r3[3:4]),]
#   Day hour Case.x Time.x
#1 Mon   10     15     18
#2 Mon   11     17    167
#3 Mon   12     20    220
#4 Mon   13     25    135
#5 Mon   14     14    167
#6 Sat    7      2     35
#7 Sun    8     18    270

data

df1 <- structure(list(Day = c("Sat", "Sun", "Sun", "Mon", "Mon", "Mon", 
"Mon", "Mon"), hour = 7:14, Case = c(2L, 8L, 10L, 15L, 17L, 20L, 
25L, 14L), Time = c(35L, 125L, 145L, 18L, 167L, 220L, 135L, 167L
)), .Names = c("Day", "hour", "Case", "Time"), class = "data.frame", 
row.names = c(NA, -8L))

Answer 2

Building on what you have

ind<-with(TestData,Day == 'Sat' |Day == 'Sun')
s<-aggregate(.~Day,data=TestData[ind,],sum)
rbind(s,TestData[!ind,])

Day hour Case Time
1 Sat    7    2   35
2 Sun   17   18  270
4 Mon   10   15   18
5 Mon   11   17  167
6 Mon   12   20  220
7 Mon   13   25  135
8 Mon   14   14  167

However, from the desired output in the question you may wish to do

s$hour<-with(TestData[ind,],hour[!duplicated(Day)])

before the rbind to get the first hour instead of the sum of hours

Day hour Case Time
1 Sat    7    2   35
2 Sun    8   18  270
4 Mon   10   15   18
5 Mon   11   17  167
6 Mon   12   20  220
7 Mon   13   25  135
8 Mon   14   14  167