slide provides a family of general purpose “sliding window” functions. The API is purposefully very similar to purrr. The goal of these functions is usually to compute rolling averages, cumulative sums, rolling regressions, or other “window” based computations.
There are 3 core functions in slide:
slide_index() computes a rolling calculation relative to an index. If you have ever wanted to compute something like a “3 month rolling average” where the number of days in each month is irregular, you might like this function.
slide_period() is similar to
slide_index() in that it slides relative to an index, but it first breaks the index up into “time blocks”, like 2 month blocks of time, and then it slides over
.x using indices defined by those blocks.
Each of these core functions have the same variants as
purrr::map(). For example,
pslide(), along with the other combinations of these variants that you might expect from having previously used purrr.
To learn more about these three functions, read the introduction vignette.
slide is NOT yet on CRAN.
You can install the development version from GitHub with:
The classic example would be to do a moving average.
slide() handles this with a combination of the
.after arguments, which control the width of the window and the alignment.
# Moving average (Aligned right) # "The current element + 2 elements before" slide_dbl(1:5, ~mean(.x), .before = 2) #>  1.0 1.5 2.0 3.0 4.0 # Align left # "The current element + 2 elements after" slide_dbl(1:5, ~mean(.x), .after = 2) #>  2.0 3.0 4.0 4.5 5.0 # Center aligned # "The current element + 1 element before + 1 element after" slide_dbl(1:5, ~mean(.x), .before = 1, .after = 1) #>  1.5 2.0 3.0 4.0 4.5
Inf, you can do a “cumulative slide” to compute cumulative expressions. I think of this as saying “give me everything before the current element.”
.complete, you can decide whether or not
.f should be evaluated on incomplete windows. In the following example, the requested window size is 3, but the first two results are computed on windows of size 1 and 2 because partial results are allowed by default. When
.complete is set to
TRUE, the first two results are not computed.
slide() iterates over data frames in a row wise fashion. Interestingly this means the default of
slide() becomes a generic row wise iterator, with nice syntax for accessing data frame columns.
There is a vignette specifically about this.
cars <- mtcars[1:4,] slide(cars, ~.x) #> [] #> mpg cyl disp hp drat wt qsec vs am gear carb #> Mazda RX4 21 6 160 110 3.9 2.62 16.46 0 1 4 4 #> #> [] #> mpg cyl disp hp drat wt qsec vs am gear carb #> Mazda RX4 Wag 21 6 160 110 3.9 2.875 17.02 0 1 4 4 #> #> [] #> mpg cyl disp hp drat wt qsec vs am gear carb #> Datsun 710 22.8 4 108 93 3.85 2.32 18.61 1 1 4 1 #> #> [] #> mpg cyl disp hp drat wt qsec vs am gear carb #> Hornet 4 Drive 21.4 6 258 110 3.08 3.215 19.44 1 0 3 1 slide_dbl(cars, ~.x$mpg + .x$drat) #>  24.90 24.90 26.65 24.48
This makes rolling regressions trivial!
library(tibble) set.seed(123) df <- tibble( y = rnorm(100), x = rnorm(100) ) # Window size of 20 rows # The current row + 19 before # (see slide_index() for how to do this relative to a date vector!) df$regressions <- slide(df, ~lm(y ~ x, data = .x), .before = 19, .complete = TRUE) df[15:25,] #> # A tibble: 11 x 3 #> y x regressions #> <dbl> <dbl> <list> #> 1 -0.556 0.519 <NULL> #> 2 1.79 0.301 <NULL> #> 3 0.498 0.106 <NULL> #> 4 -1.97 -0.641 <NULL> #> 5 0.701 -0.850 <NULL> #> 6 -0.473 -1.02 <lm> #> 7 -1.07 0.118 <lm> #> 8 -0.218 -0.947 <lm> #> 9 -1.03 -0.491 <lm> #> 10 -0.729 -0.256 <lm> #> 11 -0.625 1.84 <lm>
In many business settings, the value you want to compute is tied to some index, like a date vector. In these cases, you’ll probably want to compute sliding windows relative to the index, and not using the fixed window that
slide() provides. You can use
slide_index() to pass in both
.x and an index,
.i, and the window will be calculated relative to that index.
Here, when computing a “2 day window”, you probably don’t want
"2019-08-18" to be grouped together.
slide() has no concept of an index, so when you specify a window size of 2, it will group these two together.
slide_index(), on the other hand, will do the right thing.
x <- 1:3 i <- as.Date(c("2019-08-15", "2019-08-16", "2019-08-18")) # slide() has no concept of an "index" slide(x, ~.x, .before = 1) #> [] #>  1 #> #> [] #>  1 2 #> #> [] #>  2 3 # "index aware" slide_index(x, i, ~.x, .before = 1) #> [] #>  1 #> #> [] #>  1 2 #> #> [] #>  3
Essentially what happens is that when we get to
"2019-08-18", it “looks backwards” 1 day to set a window boundary at
"2019-08-17". Since the date at position 2,
"2019-08-16", is before
"2019-08-17", it is not included.
Powerfully, you can pass through any object to
.before that computes a value from
.i - .before. This means that you could also have used a lubridate period object (which gets even more interesting when you use
slide_period() is different from
slide_index() in that it first breaks the index into “time blocks” and then slides over
.x relative to those blocks. For example, in the monthly period slide below,
i is broken up into 4 time blocks of “the current block of monthly data, plus one block before this one”. The locations of those blocks are the locations that are used to slice
i <- as.Date(c( "2019-01-29", "2019-01-30", "2019-02-05", "2019-04-01", "2019-05-10" )) slide_period(i, i, "month", ~.x, .before = 1) #> [] #>  "2019-01-29" "2019-01-30" #> #> [] #>  "2019-01-29" "2019-01-30" "2019-02-05" #> #> [] #>  "2019-04-01" #> #> [] #>  "2019-04-01" "2019-05-10"
One neat thing to notice is that
slide_period() is aware of the distance between elements of
.i in the period you specify. The practical implication of this is that in the above example, group 3 with
2019-04-01 did not include
2019-02-05 in it, because it is more than 1 month group away.
This package is inspired heavily by SQL’s window functions. The API is similar, but more general because you can iterate over any kind of R object.
There have been multiple attempts at creating sliding window functions (I personally created
rollify(), and worked a little bit on
tsibble::slide() with Earo Wang).
I believe that slide is the next iteration of these. There are a few reasons for this:
To me, the API is more intuitive, and is more flexible because
.after let you completely control the entry point (as opposed to fixed entry points like
It is objectively faster because it is written purely in C.
slide_vec() you can return any kind of object, and are not limited to the suffixed versions:
It iterates rowwise over data frames, consistent with the vctrs framework.
I believe it is overall more consistent, backed by a theory that can always justify the sliding window generated by any combination of the parameters.
Earo and I have spoken, and we have mututally agreed that it would be best to deprecate
tsibble::slide() in favor of
Additionally, data.table’s non-equi joins have been pretty much the only solution to the problem that
slide_index() tries to solve. Their solution is robust and quite fast, and has been a nice benchmark for slide. slide is trying to solve a much narrower problem, so the API here is more focused.
In terms of performance, be aware that any specialized package that shifts the function calls to C are going to be faster than slide. For example,
RcppRoll::roll_mean() computes the rolling mean at the C level, which is bound to be faster. The purpose of slide is to be general purpose, while still being as fast as possible. This means that it can be used for more abstract things, like rolling regressions, or any other custom function that you want to use in a rolling fashion.