Help for package timeplyr

Title:

Fast Tidy Tools for Date and Date-Time Manipulation

Version:

1.1.0

Description:

A set of fast tidy functions for wrangling, completing and summarising date and date-time data. It combines 'tidyverse' syntax with the efficiency of 'data.table' and speed of 'collapse'.

License:

GPL-2 | GPL-3 [expanded from: GPL (≥ 2)]

BugReports:

https://github.com/NicChr/timeplyr/issues

Depends:

R (≥ 4.0.0)

Imports:

cheapr (≥ 1.0.0), cli, collapse (≥ 2.0.0), cppdoubles (≥ 0.2.0), data.table (≥ 1.14.8), dplyr (≥ 1.1.0), fastplyr (≥ 0.5.0), ggplot2 (≥ 3.4.0), lifecycle, lubridate (≥ 1.9.0), magrittr, pillar (≥ 1.7.0), rlang (≥ 1.0.0), scales, stringr (≥ 1.4.0), tidyselect (≥ 1.2.0), timechange (≥ 0.2.0), vctrs (≥ 0.6.0)

Suggests:

bench, knitr, nycflights13, outbreaks, rmarkdown, testthat (≥ 3.0.0), tidyr, zoo

LinkingTo:

cpp11, tzdb

Config/testthat/edition:

Encoding:

UTF-8

RoxygenNote:

7.3.2

NeedsCompilation:

yes

Packaged:

2025-03-17 18:11:59 UTC; Nmc5

Author:

Nick Christofides

[aut, cre]

Maintainer:

Nick Christofides <nick.christofides.r@gmail.com>

Repository:

CRAN

Date/Publication:

2025-03-17 18:40:02 UTC

timeplyr: Fast Tidy Tools for Date and Date-Time Manipulation

Description

A framework for handling raw date & datetime data
using tidy best-practices from the tidyverse, the efficiency of data.table, and the speed of collapse.

You can learn more about the tidyverse, data.table and collapse using the links below

tidyverse

data.table

collapse

A framework for handling raw date & datetime data
using tidy best-practices from the tidyverse, the efficiency of data.table, and the speed of collapse.

You can learn more about the tidyverse, data.table and collapse using the links below

tidyverse

data.table

collapse

Author(s)

Maintainer: Nick Christofides nick.christofides.r@gmail.com (ORCID)

Pipe operator

Description

See magrittr::%>% for details.

Usage

lhs %>% rhs

Arguments

lhs

A value or the magrittr placeholder.

rhs

A function call using the magrittr semantics.

Value

The result of calling rhs(lhs).

Time units

Description

Time units

Usage

.time_units

.period_units

.duration_units

.extra_time_units

Format

An object of class character of length 21.

An object of class character of length 7.

An object of class character of length 11.

An object of class character of length 10.

Accurate and efficient age calculation

Description

Correct calculation of ages in years using lubridate periods. Leap year calculations work as well.

Usage

age_years(start, end = if (is_date(start)) Sys.Date() else Sys.time())

age_months(start, end = if (is_date(start)) Sys.Date() else Sys.time())

Arguments

start

Start date/datetime, typically date of birth.

end

End date/datetime. Default is current date/datetime.

Value

Integer vector of age in years or months.

Create a table of common time units from a date or datetime sequence.

Description

Create a table of common time units from a date or datetime sequence.

Usage

calendar(
  x,
  label = TRUE,
  week_start = getOption("lubridate.week.start", 1),
  fiscal_start = getOption("lubridate.fiscal.start", 1),
  name = "time"
)

Arguments

x

date or datetime vector.

label

Logical. Should labelled (ordered factor) versions of week day and month be returned? Default is TRUE.

week_start

day on which week starts following ISO conventions - 1 means Monday, 7 means Sunday (default). When label = TRUE, this will be the first level of the returned factor. You can set lubridate.week.start option to control this parameter globally.

fiscal_start

Numeric indicating the starting month of a fiscal year.

name

Name of date/datetime column.

Value

An object of class tibble.

Examples

library(timeplyr)
library(lubridate)

# Create a calendar for the current year
from <- floor_date(today(), unit = "year")
to <- ceiling_date(today(), unit = "year", change_on_boundary = TRUE) - days(1)

my_seq <- time_seq(from, to, "day")
calendar(my_seq)

Get summary statistics of time delay

Description

The output is a list containing summary statistics of time delay between two date/datetime vectors. This can be especially useful in estimating reporting delay for example.

data - A data frame containing the origin, end and calculated time delay.
unit - The chosen time unit.
num - The number of time units.
summary - tibble with summary statistics.
delay - tibble containing the empirical cumulative distribution function values by time delay.
plot - A ggplot of the time delay distribution.

Usage

get_time_delay(
  data,
  origin,
  end,
  timespan = 1L,
  min_delay = -Inf,
  max_delay = Inf,
  probs = c(0.25, 0.5, 0.75, 0.95),
  .by = NULL,
  include_plot = TRUE,
  x_scales = "fixed",
  bw = "sj",
  ...
)

Arguments

data

A data frame.

origin

Origin date variable.

end

End date variable.

timespan

timespan.

min_delay

The minimum acceptable delay, all delays less than this are removed before calculation. Default is min_delay = -Inf.

max_delay

The maximum acceptable delay, all delays greater than this are removed before calculation. Default is max_delay = Inf.

probs

Probabilities used in the quantile summary. Default is probs = c(0.25, 0.5, 0.75, 0.95).

.by

(Optional). A selection of columns to group by for this operation. Columns are specified using tidy-select.

include_plot

Should a ggplot graph of delay distributions be included in the output?

x_scales

Option to control how the x-axis is displayed for multiple facets. Choices are "fixed" or "free_x".

bw

The smoothing bandwidth selector for the Kernel Density estimator. If numeric, the standard deviation of the smoothing kernel. If character, a rule to choose the bandwidth. See ?stats::bw.nrd for more details. The default has been set to "SJ" which implements the Sheather & Jones (1991) method, as recommended by the R team ?stats::density. This differs from the default implemented by stats::density() which uses Silverman's rule-of-thumb.

...

Further arguments to be passed on to ggplot2::geom_density().

Value

A list containing summary data, summary statistics and an optional ggplot.

Examples

library(timeplyr)
library(outbreaks)
library(dplyr)

ebola_linelist <- ebola_sim_clean$linelist

# Incubation period distribution

# 95% of individuals experienced an incubation period of <= 26 days
inc_distr_days <- ebola_linelist %>%
  get_time_delay(date_of_infection,
                 date_of_onset,
                 time = "days")
head(inc_distr_days$data)
inc_distr_days$unit
inc_distr_days$num
inc_distr_days$summary
head(inc_distr_days$delay) # ECDF and freq by delay
inc_distr_days$plot

# Can change bandwidth selector
inc_distr_days <- ebola_linelist %>%
  get_time_delay(date_of_infection,
                 date_of_onset,
                 time = "day",
                 bw = "nrd")
inc_distr_days$plot

# Can choose any time units
inc_distr_weeks <- ebola_linelist %>%
  get_time_delay(date_of_infection,
                 date_of_onset,
                 time = "weeks",
                 bw = "nrd")
inc_distr_weeks$plot

Rolling basic growth

Description

Calculate basic growth calculations on a rolling basis. growth() calculates the percent change between the totals of two numeric vectors when they're of equal length, otherwise the percent change between the means. rolling_growth() does the same calculation on 1 numeric vector, on a rolling basis. Pairs of windows of length n, lagged by the value specified by lag are compared in a similar manner. When lag = n then data.table::frollsum() is used, otherwise data.table::frollmean() is used.

Usage

growth(x, y, na.rm = FALSE, log = FALSE, inf_fill = NULL)

rolling_growth(
  x,
  n = 1,
  lag = n,
  na.rm = FALSE,
  partial = TRUE,
  offset = NULL,
  weights = NULL,
  inf_fill = NULL,
  log = FALSE,
  ...
)

Arguments

x

Numeric vector.

y

numeric vector

na.rm

Should missing values be removed when calculating window? Defaults to FALSE.

log

If TRUE Growth (relative change) in total and mean events will be calculated on the log-scale.

inf_fill

Numeric value to replace Inf values with. Default behaviour is to keep Inf values.

n

Rolling window size, default is 1.

lag

Lag of basic growth comparison, default is the rolling window size.

partial

Should rates be calculated outwith the window using partial windows? If TRUE (the default), (n - 1) pairs of equally-sized rolling windows are compared, their size increasing by 1 up to size n, at which point the rest of the window pairs are all of size n. If FALSE all window-pairs will be of size n.

offset

Numeric vector of values to use as offset, e.g. population sizes or exposure times.

weights

Importance weights. These can either be length 1 or the same length as x. Currently, no normalisation of weights occurs.

...

Further arguments to be passed on to frollmean.

Value

growth returns a numeric(1) and rolling_growth returns a numeric(length(x)).

Examples

library(timeplyr)

set.seed(42)
# Growth rate is 6% per day
x <- 10 * (1.06)^(0:25)

# Simple growth from one day to the next
rolling_growth(x, n = 1)

# Growth comparing rolling 3 day cumulative
rolling_growth(x, n = 3)

# Growth comparing rolling 3 day cumulative, lagged by 1 day
rolling_growth(x, n = 3, lag = 1)

# Growth comparing windows of equal size
rolling_growth(x, n = 3, partial = FALSE)

# Seven day moving average growth
roll_mean(rolling_growth(x), window = 7, partial = FALSE)

Fast Growth Rates

Description

Calculate the rate of percentage change per unit time.

Usage

growth_rate(x, na.rm = FALSE, log = FALSE, inf_fill = NULL)

Arguments

x

Numeric vector.

na.rm

Should missing values be removed when calculating window? Defaults to FALSE. When na.rm = TRUE the size of the rolling windows are adjusted to the number of non-NA values in each window.

log

If TRUE then growth rates are calculated on the log-scale.

inf_fill

Numeric value to replace Inf values with. Default behaviour is to keep Inf values.

Details

It is assumed that x is a vector of values with a corresponding time index that increases regularly with no gaps or missing values.

The output is to be interpreted as the average percent change per unit time.

For a rolling version that can calculate rates as you move through time, see roll_growth_rate.

For a more generalised method that incorporates time gaps and complex time windows, use time_roll_growth_rate.

The growth rate can also be calculated using the geometric mean of percent changes.

The below identity should always hold:

`tail(roll_growth_rate(x, window = length(x)), 1) == growth_rate(x)`

Value

numeric(1)

Examples

library(timeplyr)

set.seed(42)
initial_investment <- 100
years <- 1990:2000
# Assume a rate of 8% increase with noise
relative_increases <- 1.08 + rnorm(10, sd = 0.005)

assets <- Reduce(`*`, relative_increases, init = initial_investment, accumulate = TRUE)
assets

# Note that this is approximately 8%
growth_rate(assets)

# We can also calculate the growth rate via geometric mean

rel_diff <- exp(diff(log(assets)))
all.equal(rel_diff, relative_increases)

geometric_mean <- function(x, na.rm = TRUE, weights = NULL){
  exp(collapse::fmean(log(x), na.rm = na.rm, w = weights))
}

geometric_mean(rel_diff) == growth_rate(assets)

# Weighted growth rate

w <- c(rnorm(5)^2, rnorm(5)^4)
geometric_mean(rel_diff, weights = w)

# Rolling growth rate over the last n years
roll_growth_rate(assets)

# The same but using geometric means
exp(roll_mean(log(c(NA, rel_diff))))

# Rolling growth rate over the last 5 years
roll_growth_rate(assets, window = 5)
roll_growth_rate(assets, window = 5, partial = FALSE)

## Rolling growth rate with gaps in time

years2 <- c(1990, 1993, 1994, 1997, 1998, 2000)
assets2 <- assets[years %in% years2]

# Below does not incorporate time gaps into growth rate calculation
# But includes helpful warning
time_roll_growth_rate(assets2, window = 5, time = years2)
# Time step allows us to calculate correct rates across time gaps
time_roll_growth_rate(assets2, window = 5, time = years2, time_step = 1) # Time aware

Utility functions for checking if date or datetime

Description

Utility functions for checking if date or datetime

Usage

is_date(x)

is_datetime(x)

is_time(x)

is_time_or_num(x)

Arguments

x

Time variable.
Can be a Date, POSIXt, numeric, integer, yearmon, yearqtr, year_month or year_quarter.

Value

A logical of length 1.

Are all numbers whole numbers?

Description

Are all numbers whole numbers?

Usage

is_whole_number(x, tol = .Machine$double.eps^(2/3), na.rm = TRUE)

Arguments

x

A numeric vector.

tol

tolerance value.
The default is .Machine$double.eps^(2/3), an arbitrarily small tolerance.

na.rm

Should NA values be ignored? Default is TRUE.

Details

This is a very efficient function that returns FALSE if any number is not a whole-number and TRUE if all of them are.

Method

x is defined as a whole number vector if all numbers satisfy abs(x - round(x)) < tol.

`NA` handling

NA values are handled in a custom way.
If x is an integer, TRUE is always returned even if x has missing values.
If x has both missing values and decimal numbers, FALSE is always returned.
If x has missing values, and only whole numbers and na.rm = FALSE, then NA is returned.
Basically NA is only returned if na.rm = FALSE and x is a double vector of only whole numbers and NA values.

Inspired by the discussion in this thread: check-if-the-number-is-integer

Value

A logical vector of length 1.

Examples

library(timeplyr)
library(dplyr)

# Has built-in tolerance
sqrt(2)^2 %% 1 == 0
is_whole_number(sqrt(2)^2)

is_whole_number(1)
is_whole_number(1.2)

x1 <- c(0.02, 0:10^5)
x2 <- c(0:10^5, 0.02)

is_whole_number(x1)
is_whole_number(x2)

# Somewhat more strict than all.equal

all.equal(10^9 + 0.0001, round(10^9 + 0.0001))
is_whole_number(10^9 + 0.0001)

# Can safely be used to select whole number variables
starwars %>%
  select(where(is_whole_number))

# To reduce the size of any data frame one can use the below code

df <- starwars %>%
  mutate(across(where(is_whole_number), as.integer))

Efficient, simple and flexible ISO week calculation

Description

iso_week() is a flexible function to return formatted ISO weeks, with optional ISO year and ISO day. isoday() returns the day of the ISO week.

Usage

iso_week(x, year = TRUE, day = FALSE)

isoday(x)

Arguments

x

Date vector.

year

Logical. If TRUE then ISO Year is returned along with the ISO week.

day

Logical. If TRUE then day of the week is returned with the ISO week, starting at 1, Monday, and ending at 7, Sunday.

Value

An ISO week vector of class character.

Examples

library(timeplyr)
library(lubridate)

iso_week(today())
iso_week(today(), day = TRUE)
iso_week(today(), year = FALSE, day = TRUE)
iso_week(today(), year = FALSE, day = FALSE)

Check for missing dates between first and last date

Description

Check for missing dates between first and last date

Usage

missing_dates(x)

n_missing_dates(x)

Arguments

x

A Date or Date-Time vector.

Value

A Date vector.

Reset 'timeplyr' options

Description

Reset 'timeplyr' options

Usage

reset_timeplyr_options()

Value

Resets the timeplyr global options (prefixed with "timeplyr."):
roll_month & roll_dst.

Time resolution & granularity

Description

The definitions of resolution and granularity may evolve over time but currently the resolution defines the smallest timespan that differentiates two non-fractional instances in time. The granularity defines the smallest common time difference. A practical example would be when using dates to record data with a monthly frequency. In this case the granularity is 1 month, whereas the resolution of the data type Date is 1 day. Therefore the resolution depends only on the data type whereas the granularity depends on the frequency with which the data is recorded.

Usage

resolution(x, ...)

granularity(x, ...)

Arguments

x

Time vector.
E.g. a Date, POSIXt, numeric or any time-based vector.

...

Further arguments passed to methods.

Details

For dates and date-times, the argument exact = TRUE can be used to detect monthly/yearly granularity. In some cases this can be slow and memory-intensive so it is advised to set this to FALSE in these cases.

The default for dates is exact = TRUE whereas the default for date-times is exact = FALSE.

Value

A timespan object.

Fast rolling grouped lags and differences

Description

Inspired by 'collapse', roll_lag and roll_diff operate similarly to flag and fdiff.

Usage

roll_lag(x, n = 1L, ...)

## Default S3 method:
roll_lag(x, n = 1L, g = NULL, fill = NULL, ...)

## S3 method for class 'ts'
roll_lag(x, n = 1L, g = NULL, fill = NULL, ...)

## S3 method for class 'zoo'
roll_lag(x, n = 1L, g = NULL, fill = NULL, ...)

roll_diff(x, n = 1L, ...)

## Default S3 method:
roll_diff(x, n = 1L, g = NULL, fill = NULL, differences = 1L, ...)

## S3 method for class 'ts'
roll_diff(x, n = 1L, g = NULL, fill = NULL, differences = 1L, ...)

## S3 method for class 'zoo'
roll_diff(x, n = 1L, g = NULL, fill = NULL, differences = 1L, ...)

diff_(
  x,
  n = 1L,
  differences = 1L,
  order = NULL,
  run_lengths = NULL,
  fill = NULL
)

Arguments

x

A vector or data frame.

n

Lag. This will be recycled to match the length of x and can be negative.

...

Arguments passed onto appropriate method.

g

Grouping vector. This can be a vector, data frame or GRP object.

fill

Value to fill the first n elements.

differences

Number indicating the number of times to recursively apply the differencing algorithm. If length(n) == 1, i.e the lag is a scalar integer, an optimised method is used which avoids recursion entirely. If length(n) != 1 then simply recursion is used.

order

Optionally specify an ordering with which to apply the lags/differences. This is useful for example when applying lags chronologically using an unsorted time variable.

run_lengths

Optional integer vector of run lengths that defines the size of each lag run. For example, supplying c(5, 5) applies lags to the first 5 elements and then essentially resets the bounds and applies lags to the next 5 elements as if they were an entirely separate and standalone vector.
This is particularly useful in conjunction with the order argument to perform a by-group lag.

Details

While these may not be as fast the 'collapse' equivalents, they are adequately fast and efficient.
A key difference between roll_lag and flag is that g does not need to be sorted for the result to be correct.
Furthermore, a vector of lags can be supplied for a custom rolling lag.

roll_diff() silently returns NA when there is integer overflow. Both roll_lag() and roll_diff() apply recursively to list elements.

Value

A vector the same length as x.

Examples

library(timeplyr)

x <- 1:10

roll_lag(x) # Lag
roll_lag(x, -1) # Lead
roll_diff(x) # Lag diff
roll_diff(x, -1) # Lead diff

# Using cheapr::lag_sequence()
# Differences lagged at 5, first 5 differences are compared to x[1]
roll_diff(x, cheapr::lag_sequence(length(x), 5, partial = TRUE))

# Like diff() but x/y instead of x-y
quotient <- function(x, n = 1L){
  x / roll_lag(x, n)
}
# People often call this a growth rate
# but it's just a percentage difference
# See ?roll_growth_rate for growth rate calculations
quotient(1:10)

Fast grouped "locf" `NA` fill

Description

A fast and efficient by-group method for "last-observation-carried-forward" NA filling.

Usage

roll_na_fill(x, g = NULL, fill_limit = Inf)

Arguments

x

A vector.

g

An object use for grouping x This may be a vector or data frame for example.

fill_limit

(Optional) maximum number of consecutive NAs to fill per NA cluster. Default is Inf.

Details

Method

When supplying groups using g, this method uses radixorder(g) to specify how to loop through x, making this extremely efficient.

When x contains zero or all NA values, then x is returned with no copy made.

Value

A filled vector of x the same length as x.

Fast by-group rolling functions

Description

An efficient method for rolling sum, mean and growth rate for many groups.

Usage

roll_sum(
  x,
  window = Inf,
  g = NULL,
  partial = TRUE,
  weights = NULL,
  na.rm = TRUE,
  ...
)

roll_mean(
  x,
  window = Inf,
  g = NULL,
  partial = TRUE,
  weights = NULL,
  na.rm = TRUE,
  ...
)

roll_geometric_mean(
  x,
  window = Inf,
  g = NULL,
  partial = TRUE,
  weights = NULL,
  na.rm = TRUE,
  ...
)

roll_harmonic_mean(
  x,
  window = Inf,
  g = NULL,
  partial = TRUE,
  weights = NULL,
  na.rm = TRUE,
  ...
)

roll_growth_rate(
  x,
  window = Inf,
  g = NULL,
  partial = TRUE,
  na.rm = FALSE,
  log = FALSE,
  inf_fill = NULL
)

Arguments

x

Numeric vector, data frame, or list.

window

Rolling window size, default is Inf.

g

Grouping object passed directly to collapse::GRP(). This can for example be a vector or data frame.

partial

Should calculations be done using partial windows? Default is TRUE.

weights

Importance weights. Must be the same length as x. Currently, no normalisation of weights occurs.

na.rm

Should missing values be removed for the calculation? The default is TRUE.

...

Additional arguments passed to data.table::frollmean and data.table::frollsum.

log

For roll_growth_rate: If TRUE then growth rates are calculated on the log-scale.

inf_fill

For roll_growth_rate: Numeric value to replace Inf values with. Default behaviour is to keep Inf values.

Details

roll_sum and roll_mean support parallel computations when x is a data frame of multiple columns.
roll_geometric_mean and roll_harmonic_mean are convenience functions that utilise roll_mean.
roll_growth_rate calculates the rate of percentage change per unit time on a rolling basis.

Value

A numeric vector the same length as x when x is a vector, or a list when x is a data.frame.

Examples

library(timeplyr)

x <- 1:10
roll_sum(x) # Simple rolling total
roll_mean(x) # Simple moving average
roll_sum(x, window = 3)
roll_mean(x, window = 3)
roll_sum(x, window = 3, partial = FALSE)
roll_mean(x, window = 3, partial = FALSE)

# Plot of expected value of 'coin toss' over many flips
set.seed(42)
x <- sample(c(1, 0), 10^3, replace = TRUE)
ev <- roll_mean(x)
plot(ev)
abline(h = 0.5, lty = 2)

all.equal(roll_sum(iris$Sepal.Length, g = iris$Species),
          ave(iris$Sepal.Length, iris$Species, FUN = cumsum))
# The below is run using parallel computations where applicable
roll_sum(iris[, 1:4], window = 7, g = iris$Species)

  library(data.table)
  library(bench)
  df <- data.table(g = sample.int(10^4, 10^5, TRUE),
                   x = rnorm(10^5))
  mark(e1 = df[, mean := frollmean(x, n = 7,
                                   align = "right", na.rm = FALSE), by = "g"]$mean,
       e2 = df[, mean := roll_mean(x, window = 7, g = get("g"),
                                   partial = FALSE, na.rm = FALSE)]$mean)

Add/subtract timespans to dates and date-times

Description

A very fast method of adding time units to dates and date-times.

Usage

time_add(
  x,
  timespan,
  roll_month = getOption("timeplyr.roll_month", "xlast"),
  roll_dst = getOption("timeplyr.roll_dst", c("NA", "xfirst"))
)

time_subtract(
  x,
  timespan,
  roll_month = getOption("timeplyr.roll_month", "xlast"),
  roll_dst = getOption("timeplyr.roll_dst", c("NA", "xfirst"))
)

time_floor(x, timespan, week_start = getOption("lubridate.week.start", 1))

time_ceiling(
  x,
  timespan,
  week_start = getOption("lubridate.week.start", 1),
  change_on_boundary = is_date(x)
)

Arguments

x

Time vector.
E.g. a Date, POSIXt, numeric or any time-based vector.

timespan

timespan.

roll_month

See ?timechange::time_add. Additional choices include xlast (default) and xfirst. These work conceptually similar to skipped DST intervals.

roll_dst

See ?timechange::time_add.

week_start

change_on_boundary

?timechange::time_floor

Details

The methods are continuously being improved over time. Date arithmetic should be very fast regardless of the timespan supplied. Date-time arithmetic, specifically when supplied days, weeks, months and years, is being improved.

Value

A date, date-time, or other time-based vector.

Group by a time variable at a higher time unit

Description

time_by groups a time variable by a specified time unit like for example "days" or "weeks".
It can be used exactly like dplyr::group_by.

Usage

time_by(data, time, width = NULL, .name = NULL, .add = TRUE)

time_tbl_time_col(x)

Arguments

data

A data frame.

time

Time variable (data-masking).
E.g., a Date, POSIXt, numeric or any time variable.

width

A timespan.

.name

An optional glue specification passed to stringr::glue() which can be used to concatenate strings to the time column name or replace it.

.add

Should the time groups be added to existing groups? Default is TRUE.

x

A time_tbl_df.

Value

A time_tbl_df which for practical purposes can be treated the same way as a dplyr grouped_df.

Examples

library(dplyr)
library(timeplyr)
library(fastplyr)
library(nycflights13)
library(lubridate)

# Basic usage
hourly_flights <- flights %>%
  time_by(time_hour) # Detects time granularity

hourly_flights

monthly_flights <- flights %>%
  time_by(time_hour, "month")
weekly_flights <- flights %>%
  time_by(time_hour, "week")

monthly_flights %>%
  f_count()

weekly_flights %>%
  f_summarise(n = n(), arr_delay = mean(arr_delay, na.rm = TRUE))

# To aggregate multiple variables, use time_aggregate

flights %>%
  f_count(week = time_cut_width(time_hour, months(3)))

Cut dates and datetimes into regularly spaced date or datetime intervals

Description

Useful functions especially for when plotting time-series. time_cut makes approximately n groups of equal time range. It prioritises the highest time unit possible, making axes look less cluttered and thus prettier. time_breaks returns only the breaks.

Usage

time_cut(
  x,
  n = 5,
  timespan = NULL,
  from = NULL,
  to = NULL,
  time_floor = FALSE,
  week_start = getOption("lubridate.week.start", 1)
)

time_cut_n(
  x,
  n = 5,
  timespan = NULL,
  from = NULL,
  to = NULL,
  time_floor = FALSE,
  week_start = getOption("lubridate.week.start", 1)
)

time_cut_width(x, timespan = granularity(x), from = NULL, to = NULL)

time_breaks(
  x,
  n = 5,
  timespan = NULL,
  from = NULL,
  to = NULL,
  time_floor = FALSE,
  week_start = getOption("lubridate.week.start", 1)
)

Arguments

x

Time vector.
E.g. a Date, POSIXt, numeric or any time-based vector.

n

Number of breaks.

timespan

timespan.

from

Start time.

to

End time.

time_floor

Logical. Should the initial date/datetime be floored before building the sequence?

week_start

day on which week starts following ISO conventions - 1 means Monday (default), 7 means Sunday. This is only used when time_floor = TRUE.

Details

To retrieve regular time breaks that simply spans the range of x, use time_seq() or time_aggregate(). This can also be achieved in time_cut() by supplying n = Inf.

By default time_cut() will try to find the prettiest way of cutting the interval by trying to cut the date/date-times into groups of the highest possible time units, starting at years and ending at milliseconds.

When x is a numeric vector, time_cut will behave similar to time_cut except for 3 things:

The intervals are all right-open and of equal width.
The left value of the leftmost interval is always min(x).
Up to n breaks are created, i.e ⁠<= n⁠ breaks. This is to prioritise pretty breaks.

Value

time_breaks returns a vector of breaks.
time_cut returns either a vector or time_interval.

Examples

library(timeplyr)
library(fastplyr)
library(cheapr)
library(lubridate)
library(ggplot2)
library(dplyr)

time_cut_n(1:10, n = 5)

# Easily create custom time breaks
df <- nycflights13::flights %>%
  f_slice_sample(n = 100) %>%
  with_local_seed(.seed = 8192821) %>%
  f_select(time_hour) %>%
  fastplyr::f_arrange(time_hour) %>%
  mutate(date = as_date(time_hour))

# time_cut_n() and time_breaks() automatically find a
# suitable way to cut the data
time_cut_n(df$date) %>%
  interval_count()
# Works with datetimes as well
time_cut_n(df$time_hour, n = 5) %>%
  interval_count()
time_cut_n(df$date, timespan = "month") %>%
  interval_count()
# Just the breaks
time_breaks(df$date, n = 5, timespan = "month")

cut_dates <- time_cut_n(df$date)
date_breaks <- time_breaks(df$date)

# When n = Inf it should be equivalent to using time_cut_width
identical(time_cut_n(df$date, n = Inf, "month"),
          time_cut_width(df$date, "month"))
# To get exact breaks at regular intervals, use time_grid
weekly_breaks <- time_grid(
  df$date, "5 weeks",
  from = floor_date(min(df$date), "week", week_start = 1)
)
weekly_labels <- format(weekly_breaks, "%b-%d")
df %>%
  time_by(date, "week", .name = "date") %>%
  f_count() %>%
  mutate(date = interval_start(date)) %>%
  ggplot(aes(x = date, y = n)) +
  geom_bar(stat = "identity") +
  scale_x_date(breaks = weekly_breaks,
               labels = weekly_labels)

Time differences by any time unit

Description

The time difference between 2 date or date-time vectors.

Usage

time_diff(x, y, timespan = 1L)

Arguments

x

Start date or datetime.

y

End date or datetime.

timespan

A timespan used to divide the difference.

Value

A numeric vector recycled to the length of max(length(x), length(y)).

Examples

library(timeplyr)
library(lubridate)
time_diff(today(), today() + days(10), "days")
time_diff(today(), today() + days((0:3) * 7), weeks(1))
time_diff(today(), today() + days(100), timespan("days", 1:100))
time_diff(1, 1 + 0:100, 3)

Fast grouped time elapsed

Description

Calculate how much time has passed on a rolling or cumulative basis.

Usage

time_elapsed(
  x,
  timespan = granularity(x),
  g = NULL,
  rolling = TRUE,
  fill = NA,
  na_skip = TRUE
)

Arguments

x

Time vector.
E.g. a Date, POSIXt, numeric or any time-based vector.

timespan

timespan.

g

Object to be used for grouping x, passed onto collapse::GRP().

rolling

If TRUE (the default) then lagged time differences are calculated on a rolling basis, essentially like diff().
If FALSE then time differences compared to the index (first) time are calculated.

fill

When rolling = TRUE, this is the value that fills the first elapsed time. The default is NA.

na_skip

Should NA values be skipped? Default is TRUE.

Details

time_elapsed() is quite efficient when there are many groups, especially if your data is sorted in order of those groups. In the case that g is supplied, it is most efficient when your data is sorted by g . When na_skip is TRUE and rolling is also TRUE, NA values are simply skipped and hence the time differences between the current value and the previous non-NA value are calculated. For example, c(3, 4, 6, NA, NA, 9) becomes c(NA, 1, 2, NA, NA, 3).
When na_skip is TRUE and rolling is FALSE, time differences between the current value and the first non-NA value of the series are calculated. For example, c(NA, NA, 3, 4, 6, NA, 8) becomes c(NA, NA, 0, 1, 3, NA, 5).

Value

A numeric vector the same length as x.

Examples

library(timeplyr)
library(dplyr)
library(lubridate)

x <- time_seq(today(), length.out = 25, time = "3 days")
time_elapsed(x)
time_elapsed(x, "days", rolling = FALSE)

# Grouped example
set.seed(99)
g <- sample.int(3, 25, TRUE)

time_elapsed(x, "days", g = g)

Episodic calculation of time-since-event data

Description

This function assigns episodes to events based on a pre-defined threshold of a chosen time unit.

Usage

time_episodes(
  data,
  time,
  time_by = NULL,
  window = 1,
  roll_episode = TRUE,
  switch_on_boundary = TRUE,
  fill = 0,
  .add = FALSE,
  event = NULL,
  .by = NULL
)

Arguments

data

A data frame.

time

Date or datetime variable to use for the episode calculation. Supply the variable using tidyselect notation.

time_by

Time units used to calculate episode flags. If time_by is NULL then a heuristic will try and estimate the highest order time unit associated with the time variable. If specified, then by must be one of the three:

string, specifying either the unit or the number and unit, e.g time_by = "days" or time_by = "2 weeks"
named list of length one, the unit being the name, and the number the value of the list, e.g. list("days" = 7). For the vectorized time functions, you can supply multiple values, e.g. list("days" = 1:10).
Numeric vector. If by is a numeric vector and x is not a date/datetime, then arithmetic is used, e.g time_by = 1.

window

Single number defining the episode threshold. When rolling = TRUE events with a t_elapsed >= window since the last event are defined as a new episode.
When rolling = FALSE events with a t_elapsed >= window since the first event of the corresponding episode are defined as a new episode.
By default, window = 1 which assigns every event to a new episode.

roll_episode

Logical. Should episodes be calculated using a rolling or fixed window? If TRUE (the default), an amount of time must have passed (⁠>= window⁠) since the last event, with each new event effectively resetting the time at which you start counting.
If FALSE, the elapsed time is fixed and new episodes are defined based on how much cumulative time has passed since the first event of each episode.

switch_on_boundary

When an exact amount of time (specified in time_by) has passed, should there be an increment in ID?
The default is TRUE.
For example, if time_by = "days" and switch_on_boundary = FALSE, ⁠> 1⁠ day must have passed, otherwise ⁠>= 1⁠ day must have passed.

fill

Value to fill first time elapsed value. Only applicable when roll_episode = TRUE.
Default is 0.

.add

Should episodic variables be added to the data?
If FALSE (the default), then only the relevant variables are returned.
If TRUE, the episodic variables are added to the original data. In both cases, the order of the data is unchanged.

event

(Optional) List that encodes which rows are events, and which aren't. By default time_episodes() assumes every observation (row) is an event but this need not be the case.
event must be a named list of length 1 where the values of the list element represent the event. For example, if your events were coded as 0 and 1 in a variable named "evt" where 1 represents the event, you would supply event = list(evt = 1).

.by

(Optional). A selection of columns to group by for this operation. Columns are specified using tidyselect.

Details

time_episodes() calculates the time elapsed (rolling or fixed) between successive events, and flags these events as episodes or not based on how much time has passed.

An example of episodic analysis can include disease infections over time.

In this example, a positive test result represents an event and
a new infection represents a new episode.

It is assumed that after a pre-determined amount of time, a positive result represents a new episode of infection.

To perform simple time-since-event analysis, which means one is not interested in episodes, simply use time_elapsed() instead.

To find implicit missing gaps in time, set window to 1 and switch_on_boundary to FALSE. Any event classified as an episode in this scenario is an event following a gap in time.

The data are always sorted before calculation and then sorted back to the input order.

4 Key variables will be calculated:

ep_id - An integer variable signifying which episode each event belongs to.
Non-events are assigned NA.
ep_id is an increasing integer starting at 1. In the infections scenario, 1 are positives within the first episode of infection, 2 are positives within the second episode of infection and so on.
ep_id_new - An integer variable signifying the first instance of each new episode. This is an increasing integer where 0 signifies within-episode observations and >= 1 signifies the first instance of the respective episode.
t_elapsed - The time elapsed since the last event.
When roll_episode = FALSE, this becomes the time elapsed since the first event of the current episode. Time units are specified in the by argument.
ep_start - Start date/datetime of the episode.

data.table and collapse are used for speed and efficiency.

Value

A data.frame in the same order as it was given.

Examples

library(timeplyr)
library(dplyr)
library(nycflights13)
library(lubridate)
library(ggplot2)

# Say we want to flag origin-destination pairs
# that haven't seen departures or arrivals for a week

events <- flights %>%
  mutate(date = as_date(time_hour)) %>%
  group_by(origin, dest) %>%
  time_episodes(date, "week", window = 1)

events

episodes <- events %>%
  filter(ep_id_new > 1)
nrow(fastplyr::f_distinct(episodes, origin, dest)) # 55 origin-destinations

# As expected summer months saw the least number of
# dry-periods
episodes %>%
  ungroup() %>%
  time_by(ep_start, "week", .name = "ep_start") %>%
  count(ep_start = interval_start(ep_start)) %>%
  ggplot(aes(x = ep_start, y = n)) +
  geom_bar(stat = "identity")

A time based extension to `tidyr::complete()`.

Description

A time based extension to tidyr::complete().

Usage

time_expand(
  data,
  time = NULL,
  ...,
  .by = NULL,
  time_by = NULL,
  from = NULL,
  to = NULL,
  sort = TRUE
)

time_complete(
  data,
  time = NULL,
  ...,
  .by = NULL,
  time_by = NULL,
  from = NULL,
  to = NULL,
  sort = TRUE,
  fill = NULL
)

Arguments

data

A data frame.

time

Time variable.

...

Groups to expand.

.by

(Optional). A selection of columns to group by for this operation. Columns are specified using tidy-select.

time_by

A timespan.

from

Time series start date.

to

Time series end date.

sort

Logical. If TRUE expanded/completed variables are sorted.

fill

A named list containing value-name pairs to fill the named implicit missing values.

Details

This works much the same as tidyr::complete(), except that you can supply an additional time argument to allow for completing implicit time gaps and creating time sequences by group.

Value

A data.frame of expanded time by or across groups.

Examples

library(timeplyr)
library(dplyr)
library(lubridate)
library(nycflights13)

x <- flights$time_hour

time_num_gaps(x) # Missing hours

flights_count <- flights %>%
  fastplyr::f_count(time_hour)

# Fill in missing hours
flights_count %>%
  time_complete(time = time_hour)

# You can specify units too
flights_count %>%
  time_complete(time = time_hour, time_by = "hours")
flights_count %>%
  time_complete(time = as_date(time_hour), time_by = "days") #  Nothing to complete here

# Where time_expand() and time_complete() really shine is how fast they are with groups
flights %>%
  group_by(origin, dest) %>%
  time_expand(time = time_hour, time_by = dweeks(1))

Gaps in a regular time sequence

Description

time_gaps() checks for implicit missing gaps in time for any regular date or datetime sequence.

Usage

time_gaps(
  x,
  timespan = granularity(x),
  g = NULL,
  use.g.names = TRUE,
  check_time_regular = FALSE
)

time_num_gaps(
  x,
  timespan = granularity(x),
  g = NULL,
  use.g.names = TRUE,
  na.rm = TRUE,
  check_time_regular = FALSE
)

time_has_gaps(
  x,
  timespan = granularity(x),
  g = NULL,
  use.g.names = TRUE,
  na.rm = TRUE,
  check_time_regular = FALSE
)

Arguments

x

Time vector.
E.g. a Date, POSIXt, numeric or any time-based vector.

timespan

timespan.

g

Grouping object passed directly to collapse::GRP(). This can for example be a vector or data frame.

use.g.names

Should the result include group names? Default is TRUE.

check_time_regular

Should the time vector be checked to see if it is regular (with or without gaps)? Default is FALSE.

na.rm

Should NA values be removed? Default is TRUE.

Details

When check_time_regular is TRUE, x is passed to time_is_regular, which checks that the time elapsed between successive values are in increasing order and are whole numbers. For more strict checks, see ?time_is_regular.

Value

time_gaps returns a vector of time gaps.
time_num_gaps returns the number of time gaps.
time_has_gaps returns a logical(1) of whether there are gaps.

Examples


library(timeplyr)
library(fastplyr)
library(lubridate)
library(nycflights13)
missing_dates(flights$time_hour)
time_has_gaps(flights$time_hour)
time_num_gaps(flights$time_hour)
length(time_gaps(flights$time_hour))
time_num_gaps(flights$time_hour, g = flights$origin)

# Number of missing hours by origin and dest
flights %>%
  f_group_by(origin, dest) %>%
  f_summarise(n_missing = time_num_gaps(time_hour, "hours"))

Quick time-series ggplot

Description

time_ggplot() is a neat way to quickly plot aggregate time-series data.

Usage

time_ggplot(
  data,
  time,
  value,
  group = NULL,
  facet = FALSE,
  geom = ggplot2::geom_line,
  ...
)

Arguments

data

A data frame

time

Time variable using tidyselect.

value

Value variable using tidyselect.

group

(Optional) Group variable using tidyselect.

facet

When groups are supplied, should multi-series be plotted separately or on the same plot? Default is FALSE, or together.

geom

ggplot2 'geom' type. Default is geom_line().

...

Further arguments passed to the chosen 'geom'.

Value

A ggplot.

Examples

library(dplyr)
library(timeplyr)
library(ggplot2)
library(lubridate)

# It's as easy as this
AirPassengers %>%
  ts_as_tbl() %>%
  time_ggplot(time, value)

# And this
EuStockMarkets %>%
  ts_as_tbl() %>%
  time_ggplot(time, value, group)

# Converting this to monthly averages

EuStockMarkets %>%
  ts_as_tbl() %>%
  mutate(month = year_month_decimal(time)) %>%
  summarise(avg = mean(value),
            .by = c(group, month)) %>%
  time_ggplot(month, avg, group)

# zoo example
x.Date <- as.Date("2003-02-01") + c(1, 3, 7, 9, 14) - 1
x <- zoo::zoo(rnorm(5), x.Date)
x %>%
  ts_as_tbl() %>%
  time_ggplot(time, value)

Vector date and datetime functions

Description

These are atomic vector-based functions of the tidy equivalents which all have a "v" suffix to denote this. These are more geared towards programmers and allow for working with date and datetime vectors.

Usage

time_grid(x, timespan = granularity(x), from = NULL, to = NULL)

time_complete_missing(x, timespan = granularity(x))

time_grid_size(x, timespan = granularity(x), from = NULL, to = NULL)

Arguments

x

Time vector.
E.g. a Date, POSIXt, numeric or any time-based vector.

timespan

timespan.

from

Start time.

to

End time.

Value

Vectors (typically the same class as x) of varying lengths depending on the arguments supplied.

Examples

library(timeplyr)
library(dplyr)
library(lubridate)
library(nycflights13)
x <- unique(flights$time_hour)

# Number of missing hours
time_num_gaps(x)

# Same as above
time_grid_size(x) - length(unique(x))

# Time sequence that spans the data
length(time_grid(x)) # Automatically detects hour granularity
time_grid(x, "month")
time_grid(x, from = floor_date(min(x), "month"), to = today(),
          timespan = timespan("month"))

# Complete missing gaps in time using time_complete
y <- time_complete_missing(x, "hour")
identical(y[!y %in% x], time_gaps(x))

# Summarise time into higher intervals
quarters <- time_cut_width(y, "quarter")
interval_count(quarters)

Time ID

Description

Generate a time ID that signifies how many time steps away a time value is from the starting time point or more intuitively, this is the time passed since the first time point.

Usage

time_id(x, timespan = granularity(x), g = NULL, na_skip = TRUE, shift = 1L)

Arguments

x

Time vector.
E.g. a Date, POSIXt, numeric or any time-based vector.

timespan

timespan.

g

Object used for grouping x. This can for example be a vector or data frame. g is passed directly to collapse::GRP().

na_skip

Should NA values be skipped? Default is TRUE.

shift

Value used to shift the time IDs. Typically this is 1 to ensure the IDs start at 1 but can be 0 or even negative if for example your time values are going backwards in time.

Details

This is heavily inspired by collapse::timeid but differs in 3 ways:

The time steps need not be the greatest common divisor of successive differences
The starting time point may not necessarily be the earliest chronologically and thus time_id can generate negative IDs.
g can be supplied to calculate IDs by group.

time_id(c(3, 2, 1)) is not the same as collapse::timeid(c(3, 2, 1)). In general time_id(sort(x)) should be equal to collapse::timeid(sort(x)). The time difference GCD is always calculated using all the data and not by-group.

Value

An integer vector the same length as x.

S3-based Time Intervals (Currently very experimental and so subject to change)

Description

Inspired by both 'lubridate' and 'ivs', time_interval objects are lightweight S3 objects of a fixed width. This enables fast and flexible representation of time data such as months, weeks, and more. They are all left closed, right open intervals.

Usage

time_interval(start = integer(), width = resolution(start))

is_time_interval(x)

new_time_interval(start, width)

interval_start(x)

interval_end(x)

interval_width(x)

interval_count(x)

interval_range(x)

Arguments

start

Start time.
E.g a Date, POSIXt, numeric and more.

width

Interval width supplied as a timespan. By default this is the resolution of a time vector so for example, a date's resolution is exactly 1 day, therefore time_interval(Sys.Date()) simply represents today's date as an interval.

x

A time_interval.

Details

Currently because of limitations with the S3/S4 system, one can't use time intervals directly with lubridate periods. To navigate around this, timeplyr::timespan() can be used. e.g. instead of interval / weeks(3), use interval / timespan(weeks(3)) or even interval / "3 weeks". where interval is a time_interval.

To perform interval algebra it is advised to use the 'ivs' package. To convert a time_interval into an ivs_iv, use ivs::iv(interval_start(x), interval_end(x)).

Value

An object of class time_interval.
is_time_interval returns a logical of length 1.
interval_start returns the start times.
interval_end returns the end times.
interval_width returns the width of the interval as a timespan.
interval_count returns a data frame of unique intervals and their counts.
interval_range returns a the range of the interval.
new_time_interval is a bare-bones version of time_interval() that performs no checks.

Examples

library(dplyr)
library(timeplyr)
library(lubridate)
x <- 1:10
int <- time_interval(x, 100)
int

month_start <- floor_date(today(), unit = "months")
month_int <- time_interval(month_start, "month")
month_int

interval_start(month_int)
interval_end(month_int)

# Divide an interval into different time units
time_interval(today(), years(10)) / timespan("year")

# Cutting Sepal Length into blocks of width 1
int <- time_cut_width(iris$Sepal.Length, 1)
interval_count(int)

Is time a regular sequence? (Experimental)

Description

This function is a fast way to check if a time vector is a regular sequence, possibly for many groups. Regular in this context means that the lagged time differences are a whole multiple of the specified time unit.
This means x can be a regular sequence with or without gaps in time.

Usage

time_is_regular(
  x,
  timespan = granularity(x),
  g = NULL,
  use.g.names = TRUE,
  na.rm = TRUE,
  allow_gaps = FALSE,
  allow_dups = FALSE
)

Arguments

x

Time vector.
E.g. a Date, POSIXt, numeric or any time-based vector.

timespan

timespan.

g

Grouping object passed directly to collapse::GRP(). This can for example be a vector or data frame.
Note that when g is supplied the output is a logical with length matching the number of unique groups.

use.g.names

Should the result include group names? Default is TRUE.

na.rm

Should NA values be removed before calculation? Default is TRUE.

allow_gaps

Should gaps be allowed? Default is FALSE.

allow_dups

Should duplicates be allowed? Default is FALSE.

Value

A logical vector the same length as the number of supplied groups.

Examples

library(timeplyr)
library(lubridate)
library(dplyr)

x <- 1:5
y <- c(1, 1, 2, 3, 5)

# No duplicates or gaps allowed by default
time_is_regular(x)
time_is_regular(y)

increment <- 1

# duplicates and gaps allowed
time_is_regular(x, increment, allow_dups = TRUE, allow_gaps = TRUE)
time_is_regular(y, increment, allow_dups = TRUE, allow_gaps = TRUE)

# No gaps allowed
time_is_regular(x, increment, allow_dups = TRUE, allow_gaps = FALSE)
time_is_regular(y, increment, allow_dups = TRUE, allow_gaps = FALSE)

# Grouped
eu_stock <- ts_as_tbl(EuStockMarkets)
eu_stock <- eu_stock %>%
  mutate(date = as_date(
    date_decimal(time)
  ))

time_is_regular(eu_stock$date, g = eu_stock$group, timespan = 1,
                allow_gaps = TRUE)
# This makes sense as no trading occurs on weekends and holidays
time_is_regular(eu_stock$date, g = eu_stock$group,
                timespan = 1,
                allow_gaps = FALSE)

Fast time-based by-group rolling sum/mean - Currently experimental

Description

time_roll_sum and time_roll_mean are efficient methods for calculating a rolling sum and mean respectively given many groups and with respect to a date or datetime time index.
It is always aligned "right".
time_roll_window splits x into windows based on the index.
time_roll_window_size returns the window sizes for all indices of x.
time_roll_apply is a generic function that applies any function on a rolling basis with respect to a time index.

time_roll_growth_rate can efficiently calculate by-group rolling growth rates with respect to a date/datetime index.

Usage

time_roll_sum(
  x,
  window = timespan(Inf),
  time = NULL,
  weights = NULL,
  g = NULL,
  partial = TRUE,
  close_left_boundary = FALSE,
  na.rm = TRUE,
  ...
)

time_roll_mean(
  x,
  window = timespan(Inf),
  time = NULL,
  weights = NULL,
  g = NULL,
  partial = TRUE,
  close_left_boundary = FALSE,
  na.rm = TRUE,
  ...
)

time_roll_growth_rate(
  x,
  window = timespan(Inf),
  time = NULL,
  time_step = NULL,
  g = NULL,
  partial = TRUE,
  close_left_boundary = FALSE,
  na.rm = TRUE
)

time_roll_window_size(
  time,
  window = timespan(Inf),
  g = NULL,
  partial = TRUE,
  close_left_boundary = FALSE
)

time_roll_window(
  x,
  window = timespan(Inf),
  time = NULL,
  g = NULL,
  partial = TRUE,
  close_left_boundary = FALSE
)

time_roll_apply(
  x,
  window = timespan(Inf),
  fun,
  time = NULL,
  g = NULL,
  partial = TRUE,
  unlist = FALSE,
  close_left_boundary = FALSE
)

Arguments

x

Numeric vector.

window

Time window size as a timespan.

time

(Optional) time index.
Can be a Date, POSIXt, numeric, integer, yearmon, or yearqtr vector.

weights

Importance weights. Must be the same length as x. Currently, no normalisation of weights occurs.

g

Grouping object passed directly to collapse::GRP(). This can for example be a vector or data frame.

partial

Should calculations be done using partial windows? Default is TRUE.

close_left_boundary

Should the left boundary be closed? For example, if you specify window = "day" and time = c(today(), today() + 1),
a value of FALSE would result in the window vector c(1, 1) whereas a value of TRUE would result in the window vector c(1, 2).

na.rm

Should missing values be removed for the calculation? The default is TRUE.

...

Additional arguments passed to data.table::frollmean and data.table::frollsum.

time_step

An optional but important argument that follows the same input rules as window.
It is currently only used only in time_roll_growth_rate.
If this is supplied, the time differences across gaps in time are incorporated into the growth rate calculation. See details for more info.

fun

A function.

unlist

Should the output of time_roll_apply be unlisted with unlist? Default is FALSE.

Details

It is much faster if your data are already sorted such that !is.unsorted(order(g, x)) is TRUE.

Growth rates

For growth rates across time, one can use time_step to incorporate gaps in time into the calculation.

For example:
x <- c(10, 20)
t <- c(1, 10)
k <- Inf
time_roll_growth_rate(x, time = t, window = k) = c(1, 2) whereas
time_roll_growth_rate(x, time = t, window = k, time_step = 1) = c(1, 1.08)
The first is a doubling from 10 to 20, whereas the second implies a growth of 8% for each time step from 1 to 10.
This allows us for example to calculate daily growth rates over the last x months, even with missing days.

Value

A vector the same length as time.

Examples

library(timeplyr)
library(lubridate)
library(dplyr)
library(fastplyr)

time <- time_seq(today(), today() + weeks(3), "3 days")
set.seed(99)
x <- sample.int(length(time))

roll_mean(x, window = 7)
roll_sum(x, window = 7)

time_roll_mean(x, window = days(7), time = time)
time_roll_sum(x, window = days(7), time = time)

# Alternatively and more verbosely
x_chunks <- time_roll_window(x, window = 7, time = time)
x_chunks
vapply(x_chunks, mean, 0)

# Interval (x - 3 x]
time_roll_sum(x, window = days(3), time = time)

# An example with an irregular time series

t <- today() + days(sort(sample(1:30, 20, TRUE)))
time_elapsed(t, days(1)) # See the irregular elapsed time
x <- rpois(length(t), 10)

new_tbl(x, t) %>%
  mutate(sum = time_roll_sum(x, time = t, window = days(3))) %>%
  time_ggplot(t, sum)


### Rolling mean example with many time series

# Sparse time with duplicates
index <- sort(sample(seq(now(), now() + dyears(3), by = "333 hours"),
                     250, TRUE))
x <- matrix(rnorm(length(index) * 10^3),
            ncol = 10^3, nrow = length(index),
            byrow = FALSE)

zoo_ts <- zoo::zoo(x, order.by = index)

# Normally you might attempt something like this
apply(x, 2,
      function(x){
        time_roll_mean(x, window = dmonths(1), time = index)
      }
)
# Unfortunately this is too slow and inefficient


# Instead we can pivot it longer and code each series as a separate group
tbl <- ts_as_tbl(zoo_ts)

tbl %>%
  mutate(monthly_mean = time_roll_mean(value, window = dmonths(1),
                                       time = time, g = group))

Time based version of `base::seq()`

Description

Time based version of base::seq()

Usage

time_seq(
  from = NULL,
  to = NULL,
  time_by = NULL,
  length.out = NULL,
  roll_month = getOption("timeplyr.roll_month", "xlast"),
  roll_dst = getOption("timeplyr.roll_dst", c("NA", "xfirst"))
)

time_seq_sizes(from, to, timespan)

time_seq_v(
  from,
  to,
  timespan,
  roll_month = getOption("timeplyr.roll_month", "xlast"),
  roll_dst = getOption("timeplyr.roll_dst", c("NA", "xfirst"))
)

time_seq_v2(
  sizes,
  from,
  timespan,
  roll_month = getOption("timeplyr.roll_month", "xlast"),
  roll_dst = getOption("timeplyr.roll_dst", c("NA", "xfirst"))
)

Arguments

from

Start time.

to

End time.

time_by

A timespan. This argument may be renamed in the future.

length.out

Length of the sequence.

roll_month

Control how impossible dates are handled when month or year arithmetic is involved. Options are "preday", "boundary", "postday", "full" and "NA". See ?timechange::time_add for more details.

roll_dst

See ?timechange::time_add for the full list of details.

timespan

timespan.

sizes

Time sequence sizes.

Details

This works like seq(), but using timechange for the period calculations and base::seq.POSIXT() for the duration calculations. In many ways it is improved over seq as dates and/or datetimes can be supplied with no errors to the start and end points. Examples like,
time_seq(now(), length.out = 10, by = "0.5 days", seq_type = "dur") and
time_seq(today(), length.out = 10, by = "0.5 days", seq_type = "dur")
produce more expected results compared to
seq(now(), length.out = 10, by = "0.5 days") or
seq(today(), length.out = 10, by = "0.5 days").

For a vectorized implementation with multiple start/end times, use time_seq_v()/time_seq_v2()

time_seq_sizes() is a convenience function to calculate time sequence lengths, given start/end times.

Value

time_seq returns a time sequence.
time_seq_sizes returns an integer vector of sequence sizes.
time_seq_v returns time sequences.
time_seq_v2 also returns time sequences.

Examples

library(timeplyr)
library(lubridate)

# Dates
today <- today()
now <- now()

time_seq(today, today + months(1), time = "day")
time_seq(today, length.out = 10, time = "day")
time_seq(today, length.out = 10, time = "hour")

time_seq(today, today + months(1), time = timespan("days", 1)) # Alternative
time_seq(today, today + years(1), time = "week")
time_seq(today, today + years(1), time = "fortnight")
time_seq(today, today + years(1), time = "year")
time_seq(today, today + years(10), time = "year")
time_seq(today, today + years(100), time = "decade")

# Datetimes
time_seq(now, now + weeks(1), time = "12 hours")
time_seq(now, now + weeks(1), time = "day")
time_seq(now, now + years(1), time = "week")
time_seq(now, now + years(1), time = "fortnight")
time_seq(now, now + years(1), time = "year")
time_seq(now, now + years(10), time = "year")
time_seq(now, today + years(100), time = "decade")

# You can seamlessly mix dates and datetimes with no errors.
time_seq(now, today + days(3), time = "day")
time_seq(now, today + days(3), time = "hour")
time_seq(today, now + days(3), time = "day")
time_seq(today, now + days(3), time = "hour")

# Choose between durations or periods

start <- dmy(31012020)
# If time_type is left as is,
# periods are used for days, weeks, months and years.
time_seq(start, time = months(1), length.out = 12)
time_seq(start, time = dmonths(1), length.out = 12)
# Notice how strange base R version is.
seq(start, by = "month", length.out = 12)

# Roll forward or backward impossible dates

leap <- dmy(29022020) # Leap day
end <- dmy(01032021)
# 3 different options
time_seq(leap, to = end, time = "year",
         roll_month = "NA")
time_seq(leap, to = end, time = "year",
         roll_month = "postday")
time_seq(leap, to = end, time = "year",
         roll_month = getOption("timeplyr.roll_month", "xlast"))

Generate a unique identifier for a regular time sequence with gaps

Description

A unique identifier is created every time a specified amount of time has passed, or in the case of regular sequences, when there is a gap in time.

Usage

time_seq_id(
  x,
  timespan = granularity(x),
  threshold = 1,
  g = NULL,
  na_skip = TRUE,
  rolling = TRUE,
  switch_on_boundary = FALSE
)

Arguments

x

Time vector.
E.g. a Date, POSIXt, numeric or any time-based vector.

timespan

timespan.

threshold

Threshold such that when the time elapsed exceeds this, the sequence ID is incremented by 1. For example, if timespan = "days" and threshold = 2, then when 2 days have passed, a new ID is created. Furthermore, threshold generally need not be supplied as
timespan = "3 days" & threshold = 1
is identical to
timespan = "days" & threshold = 3.

g

Object used for grouping x. This can for example be a vector or data frame. g is passed directly to collapse::GRP().

na_skip

Should NA values be skipped? Default is TRUE.

rolling

When this is FALSE, a new ID is created every time a cumulative amount of time has passed. Once that amount of time has passed, a new ID is created, the clock "resets" and we start counting from that point.

switch_on_boundary

When an exact amount of time (specified in time_by) has passed, should there an increment in ID? The default is FALSE. For example, if time_by = "days" and switch_on_boundary = FALSE, > 1 day must have passed, otherwise >= 1 day must have passed.

Details

time_seq_id() Assumes x is regular and in ascending or descending order. To check this condition formally, use time_is_regular().

Value

An integer vector of length(x).

Examples

library(dplyr)
library(timeplyr)
library(lubridate)

# Weekly sequence, with 2 gaps in between
x <- time_seq(today(), length.out = 10, time = "week")
x <- x[-c(3, 7)]
# A new ID when more than a week has passed since the last time point
time_seq_id(x)
# A new ID when >= 2 weeks has passed since the last time point
time_seq_id(x, threshold = 2, switch_on_boundary = TRUE)
# A new ID when at least 4 cumulative weeks have passed
time_seq_id(x, timespan = "4 weeks",
            switch_on_boundary = TRUE, rolling = FALSE)
# A new ID when more than 4 cumulative weeks have passed
time_seq_id(x, timespan = "4 weeks",
            switch_on_boundary = FALSE, rolling = FALSE)

Timespans

Description

Timespans

Usage

timespan(units, num = 1L, ...)

new_timespan(units, num = 1L)

is_timespan(x)

timespan_unit(x)

timespan_num(x)

Arguments

units

A unit of time, e.g. "days", "3 weeks", lubridate::weeks(3), or just a numeric vector.

num

Number of units. E.g. units = "days" and num = 3 produces a timespan width of 3 days.

...

Further arguments passed onto methods.

x

A timespan.

Details

timespan() can be used to create objects of class 'timespan' which are used widely in timeplyr.

new_timespan() is a bare-bones version that does no checking or string parsing and is intended for fast timespan creation.

timespan_unit() is a helper that extracts the unit of time of the timespan.

timespan_num() is a helper that extracts the number of units of time.

Value

A timespan object.

Examples

library(timeplyr)

timespan("week")
timespan("day")
timespan("decade")

# Multiple units of time

timespan("10 weeks")
timespan("1.5 hours")

# These are all equivalent
timespan(NULL, 3);timespan(3);timespan(NA_character_, 3)

Additional ggplot2 scales

Description

Additional scales and transforms for use with year_months and year_quarters in ggplot2.

Usage

transform_year_month()

transform_year_quarter()

scale_x_year_month(...)

scale_x_year_quarter(...)

scale_y_year_month(...)

scale_y_year_quarter(...)

Arguments

...

Arguments passed to scale_x_continuous and scale_y_continuous.

Value

A ggplot2 scale or transform.

Turn `ts` into a `tibble`

Description

While a method already exists in the tibble package, this method works differently in 2 ways:

The time variable associated with the time-series is also returned.
The returned tibble is always in long format, even when the time-series is multivariate.

Usage

ts_as_tbl(x, name = "time", value = "value", group = "group")

## Default S3 method:
ts_as_tbl(x, name = "time", value = "value", group = "group")

## S3 method for class 'mts'
ts_as_tbl(x, name = "time", value = "value", group = "group")

## S3 method for class 'xts'
ts_as_tbl(x, name = "time", value = "value", group = "group")

## S3 method for class 'zoo'
ts_as_tbl(x, name = "time", value = "value", group = "group")

## S3 method for class 'timeSeries'
ts_as_tbl(x, name = "time", value = "value", group = "group")

Arguments

x

An object of class ts, mts, zoo, xts or timeSeries.

name

Name of the output time column.

value

Name of the output value column.

group

Name of the output group column when there are multiple series.

Value

A 2-column tibble containing the time index and values for each time index. In the case where there are multiple series, this becomes a 3-column tibble with an additional "group" column added.

Examples

library(timeplyr)
library(ggplot2)
library(dplyr)

# Using the examples from ?ts

# Univariate
uts <- ts(cumsum(1 + round(rnorm(100), 2)),
          start = c(1954, 7), frequency = 12)
uts_tbl <- ts_as_tbl(uts)

## Multivariate
mts <- ts(matrix(rnorm(300), 100, 3), start = c(1961, 1), frequency = 12)
mts_tbl <- ts_as_tbl(mts)

uts_tbl %>%
  time_ggplot(time, value)

mts_tbl %>%
  time_ggplot(time, value, group, facet = TRUE)

# zoo example
x.Date <- as.Date("2003-02-01") + c(1, 3, 7, 9, 14) - 1
x <- zoo::zoo(rnorm(5), x.Date)
ts_as_tbl(x)
x <- zoo::zoo(matrix(1:12, 4, 3), as.Date("2003-01-01") + 0:3)
ts_as_tbl(x)

Fast methods for creating year-months and year-quarters

Description

These are experimental methods for working with year-months and year-quarters inspired by 'zoo' and 'tsibble'.

Usage

year_month(x)

year_quarter(x)

YM(length = 0L)

year_month_decimal(x)

decimal_year_month(x)

YQ(length = 0L)

year_quarter_decimal(x)

decimal_year_quarter(x)

Arguments

x

A year_month, year_quarter, or any other time-based object.

length

Length of year_month or year_quarter.

Details

The biggest difference is that the underlying data is simply the number of months/quarters since epoch. This makes integer arithmetic very simple, and allows for fast sequence creation as well as fast coercion to year_month and year_quarter from numeric vectors.

Printing method is also fast.

Examples

library(timeplyr)
library(lubridate)

x <- year_month(today())

# Adding 1 adds 1 month
x + 1
# Adding 12 adds 1 year
x + 12
# Sequence of yearmonths
x + 0:12

# If you unclass, do the same arithmetic, and coerce back to year_month
# The result is always the same
year_month(unclass(x) + 1)
year_month(unclass(x) + 12)

# Initialise a year_month or year_quarter to the specified length
YM(0)
YQ(0)
YM(3)
YQ(3)

timeplyr: Fast Tidy Tools for Date and Date-Time Manipulation

Description

Author(s)

See Also

Pipe operator

Description

Usage

Arguments

Value

Time units

Description

Usage

Format

Accurate and efficient age calculation

Description

Usage

Arguments

Value

Create a table of common time units from a date or datetime sequence.

Description

Usage

Arguments

Value

Examples

Get summary statistics of time delay

Description

Usage

Arguments

Value

Examples

Rolling basic growth

Description

Usage

Arguments

Value

Examples

Fast Growth Rates

Description

Usage

Arguments

Details

Value

See Also

Examples

Utility functions for checking if date or datetime

Description

Usage

Arguments

Value

Are all numbers whole numbers?

Description

Usage

Arguments

Details

Method

NA handling

Value

Examples

Efficient, simple and flexible ISO week calculation

Description

Usage

Arguments

Value

Examples

Check for missing dates between first and last date

Description

Usage

Arguments

Value

Reset 'timeplyr' options

Description

Usage

Value

Time resolution & granularity

Description

Usage

Arguments

Details

Value

Fast rolling grouped lags and differences

`NA` handling

Fast grouped "locf" `NA` fill

A time based extension to `tidyr::complete()`.