Title:	Transform Univariate Time Series
Version:	0.0.1
Description:	Univariate time series operations that follow an opinionated design. The main principle of 'transx' is to keep the number of observations the same. Operations that reduce this number have to fill the observations gap.
License:	GPL-3
Imports:	rlang
Encoding:	UTF-8
LazyData:	true
RoxygenNote:	7.1.1
URL:	https://github.com/kvasilopoulos/transx
BugReports:	https://github.com/kvasilopoulos/transx/issues
Suggests:	dplyr, ggplot2, cli, testthat, knitr, DescTools, outliers, rmarkdown, mFilter, covr
VignetteBuilder:	knitr
Language:	en-US
Depends:	R (≥ 2.10)
NeedsCompilation:	no
Packaged:	2020-11-26 15:04:28 UTC; T460p
Author:	Kostas Vasilopoulos [aut, cre]
Maintainer:	Kostas Vasilopoulos <k.vasilopoulo@gmail.com>
Repository:	CRAN
Date/Publication:	2020-11-27 11:40:02 UTC

Removes measure of centrality from the series

Description

Removes the mean, the median or the mode from the series.

Usage

demean(x, na.rm = getOption("transx.na.rm"))

demedian(x, na.rm = getOption("transx.na.rm"))

demode(x, na.rm = getOption("transx.na.rm"))

Arguments

Value

Returns a vector with the same class and attributes as the input vector.

Examples

x <- c(2,5,10,20,30)
summary(x)

demean(x)
demedian(x)
demode(x)

Compute lagged differnces

Description

Returns suitably lagged and iterated difference

• diffx computes simple differences.

• rdffix computes percentage differences.

• ldiffx computes logged differences.

Usage

diffx(x, n = 1L, order = 1L, rho = 1, fill = NA)

rdiffx(x, n = 1L, order = 1L, rho = NULL, fill = NA)

ldiffx(x, n = 1L, order = 1L, rho = 1, fill = NA)

Arguments

Examples

x <- c(2, 4, 8, 20)
diffx(x)
rdiffx(x)
ldiffx(x)

Deterministic Trend

Description

Remove global deterministic trend information from the series.

• dt_lin removes the linear trend.

• dt_quad removes the quadratic trend.

• dt_poly removes the nth-degree polynomial trend.

Usage

dtrend_lin(x, bp = NULL, na.rm = getOption("transx.na.rm"))

dtrend_quad(x, bp = NULL, na.rm = getOption("transx.na.rm"))

dtrend_poly(x, degree, bp = NULL, na.rm = getOption("transx.na.rm"))

Arguments

Value

Returns a vector with the same class and attributes as the input vector.

Examples

set.seed(123)
t <- 1:20

# Linear trend
x <-  3*sin(t) + t
plotx(cbind(x, dtrend_lin(x)))

# Quadratic trend
x2 <-  3*sin(t) + t + t^2
plotx(cbind(raw = x2, quad = dtrend_quad(x2)))

# Introduce a breaking point at point = 10
xbp <- 3*sin(t) + t
xbp[10:20] <- x[10:20] + 15
plotx(cbind(raw = xbp, lin = dtrend_lin(xbp), lin_bp = dtrend_lin(xbp, bp = 10)))

Fill with "linear approximation"

Description

Fill with "linear approximation"

Usage

fill_linear(body, idx, ...)

Arguments

Value

Returns a vector with the same class and attributes as the input vector.

Examples

x <- c(5,3,2,2,5)
xlen <- length(x)
n <- 2
n <- pmin(n, xlen)
idx <- 1:n
body <- x[seq_len(xlen - n)]
fill_linear(body, idx)

Fill with "Last Observation Carried Forward"

Description

Fill with "Last Observation Carried Forward"

Usage

fill_locf(body, idx, fail = NA)

Arguments

Value

Returns a vector with the same class and attributes as the input vector.

Examples

x <- c(5,3,2,2,5)
lagx(x, n = 2, fill = fill_locf)
leadx(x, n = 2, fill = fill_locf)

lagx(x, n = 2, fill = fill_nocb)
leadx(x, n = 2, fill = fill_nocb)

Fill with "Next observation carried backwards"

Description

Fill with "Next observation carried backwards"

Usage

fill_nocb(body, idx, fail = NA)

Arguments

Value

Returns a vector with the same class and attributes as the input vector.

Examples

x <- c(5,3,2,2,5)
leadx(x, n = 2, fill = fill_locf)

xlen <- length(x)
n <- 2
n <- pmin(n, xlen)
idx <- (xlen - n + 1):xlen
body <- x[-seq_len(n)]
fill_locf(body, idx, NA)

Fill with "cubic spline interpolation"

Description

Fill with "cubic spline interpolation"

Usage

fill_spline(body, idx, ...)

Arguments

Value

Returns a vector with the same class and attributes as the input vector.

Examples

x <- c(5,3,NA,2,5)
fill_spline(x, 3)

Baxter-King Filter

Description

This function computes the cyclical component of the Baxter-King filter.

Usage

filter_bk(x, fill = NA, ...)

Arguments

Examples

unemp <- ggplot2::economics$unemploy
unemp_cycle <- filter_bk(unemp)
plotx(cbind(unemp, unemp_cycle))

Butterworth Filter

Description

This function computes the cyclical component of the Butterworth filter.

Usage

filter_bw(x, ...)

Arguments

Examples

unemp <- ggplot2::economics$unemploy
unemp_cycle <- filter_bw(unemp, freq = 10)
plotx(cbind(unemp, unemp_cycle))

Christiano-Fitzgerald Filter

Description

This function computes the cyclical component of the Christiano-Fitzgerald filter.

Usage

filter_cf(x, ...)

Arguments

Examples

unemp <- ggplot2::economics$unemploy
unemp_cycle <- filter_cf(unemp)
plotx(cbind(unemp, unemp_cycle))

Hamilton Filter

Description

This function computes the cyclical component of the Hamilton filter.

Usage

filter_hamilton(x, p = 4, horizon = 8, fill = NA)

Arguments

Value

Returns a vector with the same class and attributes as the input vector.

Examples

unemp <- ggplot2::economics$unemploy
unemp_cycle <- filter_hamilton(unemp)
plotx(cbind(unemp, unemp_cycle))

Hodrick-Prescot Filter

Description

This function computes the cyclical component of the Hodrick-Prescot filter.

Usage

filter_hp(x, ...)

Arguments

See Also

select_lambda

Examples

unemp <- ggplot2::economics$unemploy
unemp_cycle <- filter_hp(unemp, freq = select_lambda("monthly"))
plotx(cbind(unemp, unemp_cycle))

Trigonometric regression Filter

Description

This function computes the cyclical component of the trigonometric regression filter.

Usage

filter_tr(x, ...)

Arguments

Examples

unemp <- ggplot2::economics$unemploy
unemp_cycle <- filter_tr(unemp, pl=8, pu=40)
plotx(cbind(unemp, unemp_cycle))

Geometric Mean value

Description

Compute the sample geometric mean.

Usage

gmean(x, na.rm = getOption("transx.na.rm"))

Arguments

Value

Returns a vector with the same class and attributes as the input vector.

Compute lagged or leading values

Description

Find the "previous" (lagx()) or "next" (leadx()) values in a vector. Useful for comparing values behind of or ahead of the current values.

Usage

lagx(x, n = 1L, fill = NA)

leadx(x, n = 1L, fill = NA)

Arguments

Details

This functions has been taken and modified from the dplyr package, however, to reduce dependencies they are not imported.

Value

Returns a vector with the same class and attributes as the input vector.

Examples

x <- c(5,3,2,2,5)
lagx(x)
lagx(x, fill = mean)
lagx(x, fill = fill_nocb)

leadx(x)
leadx(x, fill = fill_locf)

Mode value

Description

Compute the sample median.

Usage

modex(x, na.rm = getOption("transx.na.rm"))

modex_int(x, na.rm = getOption("transx.na.rm"))

Arguments

Detect outliers with Tukey's method

Description

Usage

out_iqr(x, cutoff = 1.5, fill = NA, ...)

Arguments

Examples

out_iqr(c(0,1,3,4,20))

Detect outliers with Percentiles

Description

Usage

out_pt(x, pt_low = 0.1, pt_high = 0.9, fill = NA)

Arguments

Examples

x <- c(1, 3, -1, 5, 10, 100)
out_pt(x)

Detect outliers with zscore

Description

Usage

out_score_z(x, cutoff = 3, fill = NA, ...)

Arguments

Examples

out_score_z(c(0,0.1,2,1,3,2.5,2,.5,6,4,100))

Detect outliers Iglewicz and Hoaglin (1993) robust z-score method

Description

Usage

out_score_zrob(x, cutoff = 3.5, fill = NA, ...)

Arguments

Examples

out_score_zrob(c(0,0.1,2,1,3,2.5,2,.5,6,4,100))

Detect outliers with upper and lower threshold

Description

Usage

out_threshold(x, tlow = NULL, thigh = NULL, fill = NA)

Arguments

Value

Returns a vector with the same class and attributes as the input vector.

Examples

x <- c(1, 3, -1, 5, 10, 100)
out_threshold(x, tlow = 0, fill = 0)
out_threshold(x, thigh = 9, fill = function(x) quantile(x, 0.9))

Winsorize

Description

Replace extremely values that are defined by min and max.

Usage

out_winsorise(x, min = quantile(x, 0.05), max = quantile(x, 0.95))

out_winsorize(x, min = quantile(x, 0.05), max = quantile(x, 0.95))

Arguments

Value

Returns a vector with the same class and attributes as the input vector.

See Also

Winsorize

Examples

x <- c(1, 3, -1, 5, 10, 100)
out_winsorise(x)

Plotting wrapper around plot.default

Description

Helper function to only plot x as a line plot.

Usage

plotx(x, ...)

Arguments

nth Power Transformation

Description

Usage

pow(x, pow = NULL, modulus = FALSE)

Arguments

Value

Returns a vector with the same class and attributes as the input vector.

Examples

pow(2, 2)
pow(-2, 2)
pow(-2,2, TRUE)

Box-Cox Transformations

Description

Usage

pow_boxcox(x, lambda = NULL, lambda2 = NULL, ...)

Arguments

Value

Returns a vector with the same class and attributes as the input vector.

References

Box, G. E., & Cox, D. R. (1964). An analysis of transformations. Journal of the Royal Statistical Society. Series B (Methodological), 211-252. https://www.jstor.org/stable/2984418

Examples

set.seed(123)
x <- runif(10)
pow_boxcox(x, 3)

Manly(1971) Transformations

Description

The transformation was reported to be successful in transform unimodal skewed distribution into normal distribution, but is not quite useful for bimodal or U-shaped distribution.

Usage

pow_manly(x, lambda = NULL)

Arguments

Value

Returns a vector with the same class and attributes as the input vector.

Examples

set.seed(123)
x <- runif(10)
pow_manly(x, 3)

Tukey Transformations Transformations

Description

Usage

pow_tukey(x, lambda = NULL, ...)

Arguments

Value

Returns a vector with the same class and attributes as the input vector.

Examples

set.seed(123)
x <- runif(10)
pow_tukey(x, 2)

Yeo and Johnson(2000) Transformations

Description

Usage

pow_yj(x, lambda = NULL, ...)

Arguments

Value

Returns a vector with the same class and attributes as the input vector.

References

Yeo, I., & Johnson, R. (2000). A New Family of Power Transformations to Improve Normality or Symmetry. Biometrika, 87(4), 954-959. http://www.jstor.org/stable/2673623

Examples

set.seed(123)
x <- runif(10)
pow_yj(x, 3)

Change the base year

Description

Change the base year.

Usage

rebase(x, n = NULL)

rebase_origin(x)

Arguments

Value

Returns a vector with the same class and attributes as the input vector.

Examples

x <- 3:10

# New base  would be 5
rebase(x, 5)

# Or the origin
rebase_origin(x)

# Fro the base to be 100 or 0 then:
rebase(x, 5)*100
rebase(x, 5) - 1

nth Root Transformation

Description

• root: nth root

• root_sqrt: square root

• root_cubic: cubic root

Usage

root(x, root = NULL, modulus = FALSE)

root_sq(x, ...)

root_cubic(x, ...)

Arguments

Examples

root(4, 2)
root(-4, 2)

root(-4, 2, TRUE)

Rescale

Description

Usage

scale_range(x, to, na.rm = getOption("transx.na.rm"))

scale_minmax(x, na.rm = getOption("transx.na.rm"))

scale_unit_len(x, na.rm = getOption("transx.na.rm"))

Arguments

Details

To rescale a range between an arbitrary set of values [a, b], the formula becomes:

Value

Returns a vector with the same class and attributes as the input vector.

Examples

x <- c(10,5,1,-2)
scale_range(x, c(-1, 2))
scale_minmax(x)

Score transformation

Description

These functions calculate the scores according to:

• score_z: Normal(z) distribution

• score_mad: Mean absolute deviation

• score_t: t-distribution

• score_chi: chi-distribution

Usage

score_z(x, na.rm = getOption("transx.na.rm"))

score_mad(x, na.rm = getOption("transx.na.rm"))

score_t(x, na.rm = getOption("transx.na.rm"))

score_chisq(x, na.rm = getOption("transx.na.rm"))

Arguments

Details

Because function are known with different names:

• score_z is identical to std_mean

• score_mad is identical to std_median

Value

Returns a vector with the same class and attributes as the input vector.

See Also

scores

Examples

x <- seq(-3,3,0.5)
score_z(x)
score_mad(x)
score_t(x)

Selecting lambda

Description

Approaches to selecting lambda.

Usage

select_lambda(
  freq = c("quarterly", "annual", "monthly", "weekly"),
  type = c("rot", "ru2002")
)

Arguments

Details

Rule of thumb is from Hodrick and Prescot (1997):

• Lambda = 100*(number of periods in a year)^2

• Annual data = 100 x 1^2 = 100

• Quarterly data = 100 x 4^2 = 1,600

• Monthly data = 100 x 12^2 = 14,400

• Weekly data = 100 x 52^2 = 270,400

• Daily data = 100 x 365^2 = 13,322,500

Ravn and Uhlig (2002) state that lambda should vary by the fourth power of the frequency observation ratio;

• Lambda = 6.25 x (number of periods in a year)^4

Thus, the rescaled default values for lambda are:

• Annual data = 1600 x 1^4 = 6.25

• Quarterly data = 1600 x 4^4= 1600

• Monthly data = 1600 x 12^4= 129,600

• Weekly data = 1600 x 12^4 = 33,177,600

References

Hodrick, R. J., & Prescott, E. C. (1997). Postwar US business cycles: an empirical investigation. Journal of Money, credit, and Banking, 1-16.

Ravn, M. O., & Uhlig, H. (2002). On adjusting the Hodrick-Prescott filter for the frequency of observations. Review of economics and statistics, 84(2), 371-376.

Skewness/Kurtosis Value

Description

Compute the sample skewness/kurtosis

Usage

skewness(x, na.rm = getOption("transx.na.rm"))

kurtosis(x, na.rm = getOption("transx.na.rm"))

Arguments

Standarization

Description

Convert number of standard deviations by which the value of a raw score is above or below the mean value of what is being observed or measured.

Usage

std_mean(x, na.rm = getOption("transx.na.rm"))

std_median(x, na.rm = getOption("transx.na.rm"))

Arguments

Value

Returns a vector with the same class and attributes as the input vector.

Examples

x <- c(10,2,5,3)
std_mean(x)
scale(x)

std_median(x)