Exploratory and Diagnostic Tools for Time Series and Forecasts

Exploratory, descriptive and diagnostic tools for univariate and multivariate time series and tools to evaluate and compare forecasts: summary statistics and gaussianity tests; sample autocorrelation functions with the Ljung and Box (1978) and Monti (1994) tests; information criteria and Granger (1969) causality; unit root tests of Dickey and Fuller (1979) and Phillips and Perron (1988) ; the cointegration test of Johansen (1988) ; heteroskedasticity tests; non-linearity tests of McLeod and Li (1983) , Tsay (1986) and Brock, Dechert, Scheinkman and LeBaron (1996) ; spectral analysis; and forecast accuracy measures with the Diebold and Mariano (1995) , sign and Wilcoxon tests. Every tool is a pure function returning an object with print and plot methods; the graphical interfaces tooltest() and toolforecast() ('Tcl/Tk' windows or 'shiny' applications) give interactive access to all of them.

tsexplorer

Diego J. Pedregal (Universidad de Castilla-La Mancha). License GPL-3.

Exploratory, descriptive and diagnostic tools for univariate and multivariate time series, and tools to evaluate and compare forecasts. Every tool is a pure function returning an object with print() and plot() methods; the graphical interfaces tooltest() and toolforecast() give interactive access to all of them, either as Tcl/Tk windows (gui = "tcltk", the default, no extra packages) or as Shiny applications (gui = "shiny"). Data may have no dates or have them (ts, zoo, xts or a data.frame with a date column); with dates, the plots and the forecast tables show them.

Installation

From GitHub:

install.packages("remotes")
remotes::install_github("djpedregal/tsexplorer", subdir = "r")

Use

library(tsexplorer)
r <- vident(vdif(log(airpas), c(1, 1), c(1, 12)))
r          # table: ACF, PACF, Ljung-Box and Monti tests
plot(r)    # correlogram
adftest(log(airpas), lags = 1:4)
tooltest(airpas)                      # Tcl/Tk windows (default)
tooltest(airpas, gui = "shiny")       # same tool as a Shiny app
toolforecast(airpas, py, bands)       # py, bands: ts objects, like airpas
toolforecast(airpas, py, bands, gui = "shiny")

Tests

The tests compare the functions with reference results stored in inst/extdata/reference/*.json, computed on the example data and on simulated series (inst/extdata/reference/inputs/), with tolerance 1e-6.

Development

From this directory:

devtools::document()
devtools::test()
devtools::check()