Subtracting rows in R based on matching value - r

Question

I am trying to substract two rows in my dataset from each other:
Name Period Time Distance Load
Tim A 01:06:20 6000 680
Max A 01:06:20 5000 600
Leo A 01:06:20 5500 640
Noa A 01:06:20 6500 700
Tim B 00:04:10 500 80
Max B 00:04:10 500 50
Leo B 00:04:10 400 40
I want to subtract the Time, Distance and Load values of Period B from Period A for matching Names.
eg. Subtract row 5 (Tim, Period B) from row 1 (Tim, Period A)
The new values should be written into a new table looking like this:
Name Period Time Distance Load
Tim C 01:02:10 5500 600
Max C 01:02:10 4500 550
Leo C 01:02:10 5100 600
Noa C 01:06:20 6500 700
The real dataset contains many more rows. I tried to play around with dplyr but could not get the result I am looking for.
Thanks in advance

Answer 1

There are so many answers already that this is just a bit of fun at this stage. I think this way is nice as it uses unnest_wider():
library(dplyr)
library(tidyr)
library(purrr)
diff <- function(data) {
if(apply(data[2, -1], 1, function(x) all(is.na(x)))) {
data[1, -1]
} else {
data[1, -1] - data[2, -1]
}
}
df %>% group_by(Name) %>% nest() %>%
mutate(diff = map(data, diff)) %>% unnest_wider(diff) %>%
mutate(Period = "C") %>% select(Period, Time, Distance, Load)
# A tibble: 4 x 5
Name Period Time Distance Load
<chr> <chr> <time> <dbl> <dbl>
1 Tim C 01:02:10 5500 600
2 Max C 01:02:10 4500 550
3 Leo C 01:02:10 5100 600
4 Noa C 01:06:20 6500 700
Apart from the diff() function (which can probably be made neater and 'exclusively' tidyverse), this way is also shorter.
DATA
library(readr)
# courtesy of #MartinGal
df <- read_table2("Name Period Time Distance Load
Tim A 01:06:20 6000 680
Max A 01:06:20 5000 600
Leo A 01:06:20 5500 640
Noa A 01:06:20 6500 700
Tim B 00:04:10 500 80
Max B 00:04:10 500 50
Leo B 00:04:10 400 40")

Answer 2

You could filter on the two periods and then join them together, thus facilitating the subtraction of columns.
library(dplyr)
inner_join(filter(df, Period=="A"), filter(df, Period=="B"), by="Name") %>%
mutate(Period="C",
Time=Time.x-Time.y,
Distance=Distance.x-Distance.y,
Load=Load.x-Load.y) %>%
select(Name, Period, Time, Distance, Load)
Name Period Time Distance Load
1 Tim C 1.036111 hours 5500 600
2 Max C 1.036111 hours 4500 550
3 Leo C 1.036111 hours 5100 600

Answer 3

It's basically the same idea as #Edward. You could use dplyr and tidyr:
df %>%
pivot_wider(names_from="Period", values_from=c("Time", "Distance", "Load")) %>%
mutate(Period = "C",
Time = coalesce(Time_A - Time_B, Time_A),
Distance = coalesce(Distance_A - Distance_B, Distance_A),
Load = coalesce(Load_A - Load_B, Load_A)
) %>%
select(-matches("_\\w"))
returns
# A tibble: 4 x 5
Name Period Time Distance Load
<chr> <chr> <time> <dbl> <dbl>
1 Tim C 01:02:10 5500 600
2 Max C 01:02:10 4500 550
3 Leo C 01:02:10 5100 600
4 Noa C 01:06:20 6500 700
Data
df <- read_table2("Name Period Time Distance Load
Tim A 01:06:20 6000 680
Max A 01:06:20 5000 600
Leo A 01:06:20 5500 640
Noa A 01:06:20 6500 700
Tim B 00:04:10 500 80
Max B 00:04:10 500 50
Leo B 00:04:10 400 40")

Answer 4

Here is a different approach which groups by Name to get the difference.
library(dplyr)
library(chron)
df <- structure(list(Name = structure(c(4L, 2L, 1L, 3L, 4L, 2L, 1L), .Label = c("Leo", "Max", "Noa", "Tim"), class = "factor"),
Period = structure(c(1L,1L, 1L, 1L, 2L, 2L, 2L), .Label = c("A", "B"), class = "factor"),
Time = structure(c(2L, 2L, 2L, 2L, 1L, 1L, 1L), .Label = c("0:04:10", "1:06:20"), class = "factor"),
Distance = c(6000L, 5000L, 5500L, 6500L, 500L, 500L, 400L),
Load = c(680L, 600L, 640L, 700L, 80L, 50L, 40L)), class = "data.frame", row.names = c(NA, -7L))
df %>%
mutate(Time = times(Time)) %>%
group_by(Name) %>%
mutate(Time = lag(Time) - Time,
Distance = lag(Distance) - Distance,
Load = lag(Load) - Load,
Period = LETTERS[which(LETTERS == Period) + 1]) %>%
filter(!is.na(Time))

Answer 5

You can use data.table too.
dt <- data.table(Name = c('Tim', 'Max', 'Leo', 'Noa', 'Tim', 'Max', 'Leo'),
Period = c('A', 'A', 'A', 'A', 'B', 'B', 'B'),
Time = c('01:06:20', '01:06:20' , '01:06:20' , '01:06:20' , '00:04:10' , '00:04:10' , '00:04:10' ),
Distance = c(6000, 5000, 5500, 6500, 500, 500, 400 ),
Load = c(680, 600, 640, 700, 80, 50, 40))
Then the first thing to do is to convert the Time var:
dt[, Time := as.POSIXct(Time, format = "%H:%M:%S")]
sapply(dt, class)
Then you use dcast.data.table:
dtCast <- dcast.data.table(dt, Name ~ Period, value.var = c('Time', 'Distance', 'Load'))
And then you create a new object:
dtFinal <- dtCast[,list(Period = 'C',
Time = Time_A - Time_B,
Distance = Distance_A - Distance_B,
Load = Load_A - Load_B),
by = 'Name']
Mind that if you want to convert the Time to the same format as above, you need to do the following:
library(hms)
dtFinal[, Time := as_hms(Time)]