Applying functions to rolling windows

Manipulating Time Series Data in R

Harrison Brown

Graduate Researcher in Geography

Rolling functions from zoo

zoo::roll*:

rollmean()
rollmax()
rollsum()
...

Limitations:

No rollmin()!
Only certain functions exist

rollapply

rollapply(data = ftse,
          width = 7,
          FUN = min,
          align = 'right',
          fill = NA)

Arguments:

data: Time series object
width: Width of window (k)
FUN: Summary function (mean, max, ...)
align: Window alignment ('left', 'right', or 'center')
fill: Values to fill in outside of window

Roll-applying functions

Goal:

Find range of values within 30-day window
Approximate 'variation' in data

daily_temp

2019-01-01 13.191304
2019-01-02  6.945833
2019-01-03  8.495833
2019-01-04  8.683333
2019-01-05  7.400000
...

find_range <- function(x){
  max(x) - min(x)
}

daily_temp_range <- 
  rollapply(
      daily_temp,
    FUN = find_range,
    width = 30,
    align = 'right',
    fill = NA
  )

Roll-applying functions

autoplot(daily_temp_range) + 
  theme_light() + 
  labs(y = 'Temperature Range')

Plot of the 'daily temperature range' time series. This plot shows how the range of values changes across time; a range is calculated for a 30-day rolling window, and the difference between the minimum and maximum values is reported in the plot.

Let's practice!

Manipulating Time Series Data in R