Interface to 'JDemetra+' Seasonal Adjustment Software

Interface around 'JDemetra+' (< https://github.com/jdemetra/jdemetra-app>), the seasonal adjustment software officially recommended to the members of the European Statistical System (ESS) and the European System of Central Banks. It offers full access to all options and outputs of 'JDemetra+', including the two leading seasonal adjustment methods TRAMO/SEATS+ and X-12ARIMA/X-13ARIMA-SEATS.

RJDemetra is a R interface to JDemetra+, the seasonal adjustment software officially recommended to the members of the European Statistical System (ESS) and the European System of Central Banks. JDemetra+ implements the two leading seasonal adjustment methods TRAMO/SEATS+ and X-12ARIMA/X-13ARIMA-SEATS.

Besides seasonal adjustment, JDemetra+ bundles other time series models that are useful in the production or analysis of economic statistics, including for instance outlier detection, nowcasting, temporal disaggregation or benchmarking.

For more details on the JDemetra+ software see https://github.com/jdemetra/jdemetra-app.

RJDemetra offers full access to all options and outputs of JDemetra+.

Installation

RJDemetra relies on the rJava package and Java SE 8 or later version is required.

# Install release version from CRAN
install.packages("RJDemetra")
 
# Install development version from GitHub
# install.packages("devtools")
devtools::install_github("jdemetra/rjdemetra")

Basic example

To seasonally adjust a time series with a pre-defined specification you can either use the x13_def() function for the X-13ARIMA method or the tramoseats_def() function for the TRAMO-SEATS method.

library(RJDemetra)
myseries <- ipi_c_eu[, "FR"]
x13_model <- x13_def(myseries) # X-13ARIMA method
ts_model <- tramoseats_def(myseries) # TRAMO-SEATS method
 
# Basic plot with the original series, the trend and the SA series
plot(x13_model, type_chart = "sa-trend")

 
# S-I ratio
plot(x13_model$decomposition)

News

RJDemetra 0.1.2

Major changes

All _def functions are now deprecated and replaced by the functions with the same name but without _def. Use: x13_spec instead of x13_spec_def, x13 instead of x13_def, tramoseats_spec instead of tramoseats_spec_def, tramoseats instead of tramoseats_def, regarima_spec_tramoseats instead of regarima_spec_def_tramoseats, regarima_tramoseats instead of regarima_def_tramoseats, regarima_x13 instead of regarima_def_x13 and regarima_spec_x13 instead of regarima_spec_def_x13.
object argument renamed by spec in x13_spec, tramoseats_spec, regarima_spec_x13 and regarima_spec_tramoseats.

New functionalities

Parameter preliminary.check added to the specifications functions (regarima_spec_tramoseats, tramoseats_spec, regarima_spec_x13 and x13_spec). By default (preliminary.check = TRUE), JDemetra+ checks the quality of the input series and exclude highly problematic ones: e.g. these with a number of identical observations and/or missing values above pre-specified threshold values. When preliminary.check = FALSE, the thresholds are ignored and process is performed, when possible. (issue #39)
Error message returned when the seasonal adjustment fails due to the preliminary.check.
Possibility to use user-defined calendar regressors. To do it use tradingdays.option = "UserDefined and add new regressors variables (usrdef.varEnabled = TRUE to enable user-defined regressors and usrdef.var to define the regressors) using usrdef.varType = "Calendar".
usrdef.varType argument is recycled with the number of variables defined in the usrdef.var parameter.
News functions to only get the Java object from a seasonal adjustment or a pre-ajustment method: jx13, jtramoseats, jregarima, jregarima_x13, jregarima_tramoseats and get_jmodel. Therefore, there is no formatting and the computation is faster than the non 'j' functions (x13, tramoseats, regarima, regarima_x13, regarima_tramoseats and get_model). To manipulate these objects, there are three functions: get_dictionary to get the indicators that can be extracted, get_indicators to extract these indicators and jSA2R to get the formatted R model.