Output from X13-Arima

In this chapter

The output of X13-Arima procedure is detailed (Reg-Arima pre-processing, and X11 Decomposition).

(pre-processing or pre-adjustment or pre-treatment are used interchangeably)

This presentation is broken down by categories to make navigation easier

  • series

    • final series resulting from a complete SA process, typically final series include re-allocated pre-adjustment effects

    • pre-adjustment series

    • decomposition series (based on the linearized series without pre-adjustment effects)

  • parameters from

    • pre-processing

    • decomposition

  • diagnostics on

    • pre-processing (eg: tests on Reg-arima model residuals)

    • decomposition (M Statistics)

    • SA process: tests for residual, seasonality, residual Trading Days effects…

How to generate output ?

From Graphical User Interface

Using the Cruncher

With rjd3x13

Description details

Name: name: is name in… unless difference is signalled

Fullname and how to parametrize (?)

Series

Final series from Seasonal Adjustment Process

Name Description GUI display Fullname
y Original series Main Results > Table y
y_f Forecasts of original series Main Results > Table y_f(?)
y_ef Forecasts errors of original series Main Results > Table y_ef(?)
y_b Backcasts of original series Main Results > Table y_b(?)
y_eb Backcasts errors of original series Main Results > Table y_eb(?)
sa Seasonally adjusted series Main Results > Table sa
sa_f Forecasts of seasonally adjusted series Main Results > Table sa_f
t Trend component Main Results > Table t
t_f Forecasts of trend Main Results > Table t_f
s Seasonal component Main Results > Table s
s_f Forecasts of seasonal component Main Results > Table s_f
i Irregular component Main Results > Table i
i_f Forecasts of irregular component Main Results > Table i_f
ycal series corrected for calendar effects: y_cal = yc-/cal Pre-Processing > Pre-adjustment series ycal
ycal_f forecasts of the series corrected for calendar effects Pre-Processing > Pre-adjustment series ycal_f(?)
ycal_b backcasts of the series corrected for calendar effects Pre-Processing > Pre-adjustment series ycal_b(?)
benchmarking.original Unbenchmarked sa series No benchmarking.original
benchmarking.target Series used to compute annual targets No benchmarking.target
benchmarking.result Final benchmarked sa series No benchmarking.result

Final series with X-11 style names

Backcasts of final series can be retrieved only from this table.

GUI display: Decomposition (X11)>D-Final-Table

Name Description
finals.d11 Seasonally adjusted series (=sa)
finals.d12 Trend component (=t)
finals.d13 Irregular component (=i)
finals.d16 Seasonal component (=s)
finals.d18 Calendar effects (=cal)
finals.d11a Forecasts of seasonally adjusted series (=sa_f)
finals.d12a Forecasts of trend (=t_f)
finals.d16a Forecasts of seasonal component (=s_f)
finals.d18a Forecasts of the calendar effects (=cal_f)
finals.d11b Backcasts of seasonally adjusted series
finals.d12b Backcasts of trend
finals.d16b Backcasts seasonal component
finals.d18b Backcasts of the calendar effects (=cal_b)

Robustified final series

GUI display: Decomposition (X11)> E-Table

Name Description
finals.e1 Original series corrected for most important outliers
finals.e2 Final seasonally adjusted series corrected for most important outliers
finals.e3 Final Irregular corrected for most important outliers
finals.e11 Robust estimation of final seasonally adjusted series

Pre-processing series

GUI display: Pre-Processing > Pre-adjustment series

attention couper à la fin series not displaed in GUI ? generable in GUI output ?

Name Description Fullname
yc interpolated series. Untransformed yc
yc_f forecasts of the interpolated series yc_f(?)
yc_b backcasts of the interpolated series yc_b(?)
ylin linearized series (series without pre-processing and regression effects). l=yc-/det. Untransformed ylin
ylin_f forcasts of the linearized series. Untransformed ylin_f(?)
ylin_b backcasts of the linearized series. Untransformed ylin_b(?)
det all deterministic effects (including pre-processing, but without trend polynomial effect). Untransformed det
det_f forcasts of all deterministic effects (including pre-processing, but without trend polynomial effect). Untransformed det_f(?)
det_b backcasts of all deterministic effects (including pre-processing, but without trend polynomial effect). Untransformed det_b(?)
cal all calendar effects (including pre-processings). cal=tde+*mhe. Untransformed cal
cal_f forecasts of all calendar effects. Untransformed cal_f(?)
cal_b backcasts of all calendar effects. Untransformed cal_b(?)
ycal series corrected for calendar effects: y_cal = yc-/cal. Untransformed ycal
ycal_f forecasts of the series corrected for calendar effects. Untransformed ycal_f(?)
ycal_b backcasts of the series corrected for calendar effects. Untransformed ycal_b(?)
tde trading days effects (including leap year/length of period, including pre-processings). Untransformed tde
tde_f forecasts of the trading days effects. Untransformed tde_f(?)
tde_b backcasts of the trading days effects. Untransformed tde_b(?)
ee Easter effects. Untransformed ee
ee_f forecasts of the Easter effects. Untransformed ee_f(?)
ee_b backcasts of the Easter effects. Untransformed ee_b(?)
omhe other mothing holidays effects. Untransformed omhe
omhe_f forecasts of the other mothing holidays effects. Untransformed omhe_f(?)
omhe_b backcasts of the other mothing holidays effects. Untransformed omhe_b(?)
mhe all moving holidays effects. mhe=ee+*rmde+*omhe. Untransformed mhe
mhe_f forecats of all moving holidays effects. mhe=ee+*rmde+*omhe. Untransformed mhe_f(?)
mhe_b backcasts of all moving holidays effects. mhe=ee+*rmde+*omhe. Untransformed mhe_b(?)
out all outliers effects. Untransformed out
out_f forecasts of all outliers effects. Untransformed out_f(?)
out_b backcasts of all outliers effects. Untransformed out_b(?)
reg all other regression effects (outside outliers and calendars). Untransformed reg
reg_f forecasts of all other regression effects (outside outliers and calendars). Untransformed reg_f(?)
reg_b backcasts of all other regression effects (outside outliers and calendars). Untransformed reg_b(?)
l linearized series (transformed series without pre-processing and regression effects). Transformed) l
l_f forecasts of the linearized series. Transformed) l_f(?)
l_b backcasts of the linearized series. Transformed) l_b(?)
full_res full residuals full_res
out_t outliers effects associated to the trend out_t
out_s outliers effects associated to the seasonal out_s
out_i outliers effects associated to the irregular out_i
reg_t other regression effects associated to the trend reg_t
reg_s other regression effects associated to the seasonal reg_s
reg_i other regression effects associated to the irregular reg_i
reg_sa other regression effects associated to the sa reg_sa
reg_u other undefined regression effects (split between the components reg_u
reg_y other regression effects removed from the series (not in the components reg_y
det_t all regression effects associated to the trend det_t
det_s all regression effects associated to the seasonal det_s
det_i all regression effects associated to the irregular det_i
out_t_f outliers effects associated to the trend out_t_f(?)
out_s_f outliers effects associated to the seasonal out_s_f(?)
out_i_f outliers effects associated to the irregular out_i_f(?)
reg_t_f other regression effects associated to the trend reg_t_f(?)
reg_s_f other regression effects associated to the seasonal reg_s_f(?)
reg_i_f other regression effects associated to the irregular reg_i_f(?)
reg_sa_f other regression effects associated to the sa reg_sa_f(?)
reg_u_f other undefined regression effects (split between the components reg_u_f(?)
reg_y_f other regression effects removed from the series (not in the components reg_y_f(?)
det_t_f all regression effects associated to the trend det_t_f(?)
det_s_f all regression effects associated to the seasonal det_s_f(?)
det_i_f all regression effects associated to the irregular det_i_f(?)
out_t_b outliers effects associated to the trend out_t_b(?)
out_s_b outliers effects associated to the seasonal out_s_b(?)
out_i_b outliers effects associated to the irregular out_i_b(?)
reg_t_b other regression effects associated to the trend reg_t_b(?)
reg_s_b other regression effects associated to the seasonal reg_s_b(?)
reg_i_b other regression effects associated to the irregular reg_i_b(?)
reg_sa_b other regression effects associated to the sa reg_sa_b(?)
reg_u_b other undefined regression effects (split between the components reg_u_b(?)
reg_y_b other regression effects removed from the series (not in the components reg_y_b(?)
det_t_b all regression effects associated to the trend det_t_b(?)
det_s_b all regression effects associated to the seasonal det_s_b(?)
det_i_b all regression effects associated to the irregular det_i_b(?)

Pre-processing series in X11 A-Table

GUI display: Decomposition (X11)>A-Table

Name Description GUI dsiplay
preadjustment.a1 original series yes
preadjustment.a1a forecasts of the original series (=y_f) yes
preadjustment.a1b backcasts of the original series (=y_b) no
preadjustment.a6 trading days effects (=tde) yes
preadjustment.a7 moving holidays effects (including easter) (=tde+ee) yes
preadjustment.a8 outliers effects (=out) yes
preadjustment.a8t outliers effects associated to the trend (=out_t) yes
preadjustment.a8i outliers effects associated to the irregular (=out_i) yes
preadjustment.a8s outliers effects associated to the seasonal (=out_s) yes
preadjustment.a9 other regression effects (=reg) yes
preadjustment.a9cal other regression effects, associated to the calendar component yes
preadjustment.a9u other regression effects, split in the different components (allocation = “undefined”) yes
preadjustment.a9sa other regression effects, associated to the SA series (allocation= “sa”) yes
preadjustment.a9ser other regression effects, removed from the series and not integrated in the final components (allocation =“series”) yes
  • allocation: brief explanation and link

Diagnostics

remark: stand alone tests: - in GUI - in R (rjd3toolkit)

Tests on residuals

ONE TEST example : LB TEST, link to M chap what do we describe here

Seasonality tests

ONE TEST example : Friedman TEST, link to M chap what do we describe here

Parameters and other estimation results

char

Name Description GUI display
period number of observations per year
span.start start date of SA estimation
span.end end date of SA estimation
span.n number of observations in the original series
span.missing number of missing values in the original series
regression.espan.start start date of pre-processing model estimation Main Results and Pre-processing
regression.espan.end end date of pre-processing model estimation Main Results and Pre-processing
regression.espan.n number of observations in pre-processing model estimation Main Results and Pre-processing
regression.espan.missing number of missing values in series used for pre-processing model estimation