persephone

This package is an object oriented wrapper around the R Package RJDemetra, which performs time series adjustments with the java library JDemetra+

Installation

# install the package from GitHub
devtools::install_github("statistikat/persephone")

# optional: also install vignettes (needs prettydoc)
devtools::install_github("statistikat/persephone", build_vignettes = TRUE)

Usage

Objects can be constructed with perX13 or perTramo. Subseqentually, the run method runs the model and output gives access to the output object from RJDemetra.

library(persephone)
obj <- perX13(AirPassengers)
plot(obj)

The original output from RJDemetra can be inspected via the output field.

obj$run()
obj$output$regarima
#> y = regression model + arima (0, 1, 0, 0, 1, 1)
#> Log-transformation: yes
#> Coefficients:
#>           Estimate Std. Error
#> BTheta(1)  -0.5007      0.081
#> 
#>              Estimate Std. Error
#> Monday      -0.001527      0.004
#> Tuesday     -0.007677      0.004
#> Wednesday   -0.001125      0.004
#> Thursday    -0.005350      0.004
#> Friday       0.004676      0.004
#> Saturday     0.003025      0.004
#> Easter [1]   0.017999      0.008
#> AO (5-1951)  0.109258      0.020
#> 
#> 
#> Residual standard error: 0.03006 on 131 degrees of freedom
#> Log likelihood = 271.5, aic = 947.6 aicc = 949.5, bic(corrected for length) = -6.674

Further reading

More information can be found on the github-pages site for persephone.

• An overview of the package is available in the useR!2019 slides.
• The plotting vignette contains examples of interactive plots htat can be created with persephone.
• More information about hierarchical time series can be found in the hierarchical timeseries vignette.