I am trying to use rollapplyr() on a data table an use it to generate various columns via .SDcols. Rather than putting the results for one column next to the results for another column, however, rollapplyr() stacks the results for each column below one another. Below some code to illustrate what I mean:
library(data.table)
library(zoo)
cars <- data.table(mtcars)[, c('cyl', 'mpg', 'hp')]
setorder(cars, cyl)
maCols <- c('mpg_ma', 'hp_ma')
cars[, rollapplyr(.SD, mean, width = 3, align = 'right', partial = TRUE),
by = cyl,
.SDcols = c('mpg', 'hp')]
This code generates a data.table of two columns (cyl and V1) of 64 rows rather than a data table of three columns (cyl, V1 and V2) of 32 columns. Is there a way to obtain the latter?
You could try:
cars[, lapply(.SD, function(x) rollapplyr(x, mean, width = 3, align = 'right', partial = TRUE)),
by = cyl,
.SDcols = c('mpg', 'hp')]
Output:
cyl mpg hp
1: 4 22.80000 93.00000
2: 4 23.60000 77.50000
3: 4 23.33333 83.33333
4: 4 26.53333 74.33333
5: 4 28.53333 71.00000
6: 4 32.23333 61.00000
7: 4 28.60000 71.33333
8: 4 27.56667 76.00000
9: 4 24.93333 84.66667
10: 4 27.90000 90.00000
11: 4 25.93333 104.33333
12: 6 21.00000 110.00000
13: 6 21.00000 110.00000
14: 6 21.13333 110.00000
15: 6 20.16667 108.33333
16: 6 19.56667 112.66667
17: 6 18.36667 117.00000
18: 6 18.90000 140.33333
19: 8 18.70000 175.00000
20: 8 16.50000 210.00000
21: 8 16.46667 200.00000
22: 8 16.00000 201.66667
23: 8 16.30000 180.00000
24: 8 14.30000 188.33333
25: 8 12.00000 200.00000
26: 8 11.83333 216.66667
27: 8 13.53333 198.33333
28: 8 15.13333 176.66667
29: 8 14.66667 181.66667
30: 8 15.90000 190.00000
31: 8 16.10000 228.00000
32: 8 16.66667 258.00000