Closed AlexanderBae closed 4 years ago
Hi, This is normal: for the decomposition, the raw time-series is corrected for trading days and outliers. The "y_cmp" time series corresponds corresponds to this "linearized" series. You can find more details here: https://jdemetradocumentation.github.io/JDemetra-documentation/pages/theory/SA_lin%20—%20kopia.html. Below a simple example with no trading day adjustment and with only one additive outlier in october 2009. You can notice that the only difference with the "original" series occurs in october 2009 and it is equal to the coefficient of the outlier.
library(RJDemetra)
tsspec< - tramoseats_spec(spec = "RSA0",
usrdef.outliersEnabled = TRUE,
usrdef.outliersType = "AO",
usrdef.outliersDate = "2009-10-01")
m <- tramoseats(ipi_c_eu[, "FR"], spec = tsspec, userdefined = NULL)
m$regarima
#> y = regression model + arima (0, 1, 1, 0, 1, 1)
#> Log-transformation: no
#> Coefficients:
#> Estimate Std. Error
#> Theta(1) -0.3783 0.052
#> BTheta(1) -0.5815 0.047
#>
#> Estimate Std. Error
#> Mean 0.009894 0.030
#> AO (10-2009) -1.169685 1.463
#>
#>
#> Residual standard error: 1.969 on 318 degrees of freedom
#> Log likelihood = -679.7, aic = 1369 aicc = 1370, bic(corrected for length) = 1.426
round(ipi_c_eu[,"FR"] - m$decomposition$components[,"y_cmp"], 2)
#> Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
#> 1990 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
#> 1991 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
#> 1992 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
#> 1993 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
#> 1994 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
#> 1995 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
#> 1996 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
#> 1997 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
#> 1998 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
#> 1999 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
#> 2000 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
#> 2001 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
#> 2002 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
#> 2003 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
#> 2004 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
#> 2005 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
#> 2006 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
#> 2007 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
#> 2008 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
#> 2009 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 -1.17 0.00 0.00
#> 2010 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
#> 2011 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
#> 2012 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
#> 2013 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
#> 2014 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
#> 2015 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
#> 2016 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
#> 2017 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
If you want to get the original series used by your model, the best solution is to use the function get_ts
get_ts(m)
#> Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
#> 1990 90.5 92.6 101.9 95.2 92.1 103.3 91.8 65.5 99.0 102.8 94.3 93.1
#> 1991 90.9 89.6 99.9 93.3 88.3 103.0 89.7 65.1 98.2 100.8 95.8 93.2
#> 1992 89.4 89.0 99.5 93.0 89.1 101.3 89.4 64.1 94.9 98.6 92.2 90.5
#> 1993 85.3 84.3 93.2 87.8 83.5 95.4 86.2 60.1 92.1 95.8 88.1 88.3
#> 1994 84.9 84.0 94.1 90.1 86.8 100.4 90.8 64.5 96.8 101.0 96.6 96.3
#> 1995 90.4 90.5 100.4 94.5 89.7 103.7 93.8 65.5 99.7 101.8 94.6 98.1
#> 1996 90.3 88.8 100.7 93.8 91.2 104.4 92.3 67.2 100.2 102.3 96.9 97.2
#> 1997 90.5 91.6 104.0 99.7 93.9 108.8 98.2 73.4 105.8 111.8 102.4 105.4
#> 1998 99.2 99.0 109.4 103.0 100.7 114.8 104.9 73.3 109.6 112.7 105.9 105.1
#> 1999 100.5 98.6 111.8 104.3 101.3 117.4 106.6 74.9 113.4 118.2 110.9 109.8
#> 2000 104.8 104.9 118.9 110.2 108.0 122.5 111.8 80.5 117.5 121.7 114.3 115.5
#> 2001 108.8 109.2 123.7 111.8 108.4 124.7 111.1 84.2 117.8 121.0 111.6 109.2
#> 2002 106.6 107.0 121.4 112.8 106.4 122.2 109.7 82.3 117.1 118.7 113.0 106.4
#> 2003 105.4 105.7 120.1 111.1 102.8 118.3 108.8 78.7 115.9 119.9 110.8 107.9
#> 2004 105.8 107.0 120.0 112.1 105.8 123.6 112.0 78.4 120.0 122.0 112.0 108.4
#> 2005 109.1 106.7 117.9 113.5 106.8 122.3 110.3 80.0 121.4 118.4 115.2 109.8
#> 2006 107.3 106.3 121.9 112.5 110.8 126.7 112.5 82.5 122.2 121.9 113.7 111.7
#> 2007 109.5 110.0 123.8 114.2 112.6 127.0 115.2 85.7 121.2 124.7 115.2 111.0
#> 2008 111.4 112.2 123.0 116.8 108.4 122.1 112.6 81.9 117.3 116.3 102.4 97.8
#> 2009 91.7 90.4 100.1 93.2 91.4 105.5 96.8 71.6 104.5 104.8 99.3 92.9
#> 2010 93.7 93.5 106.8 99.9 96.9 108.9 101.7 73.2 107.2 108.2 102.5 97.9
#> 2011 100.4 102.0 112.0 103.7 102.9 111.3 105.0 76.6 108.4 109.7 106.0 98.4
#> 2012 97.8 97.6 110.2 102.3 97.0 108.8 102.5 77.5 106.6 105.2 99.3 95.7
#> 2013 94.3 95.9 106.3 102.5 96.8 108.2 100.7 74.3 104.6 106.1 101.0 95.2
#> 2014 95.0 97.2 107.4 101.7 93.6 107.7 100.6 74.3 105.5 105.3 98.5 96.8
#> 2015 95.3 96.6 108.8 101.9 96.5 110.0 99.9 76.9 107.1 108.0 101.8 97.3
#> 2016 98.5 97.9 107.4 103.4 96.9 107.8 99.6 77.5 106.2 106.5 104.2 97.1
#> 2017 97.4 97.5 112.0 103.0 100.4 111.2 103.4 79.3 109.7 114.0 107.7 101.4
Created on 2019-12-13 by the reprex package (v0.3.0)
Hi! I have a problem with seasonally adjusting several time series. In particular after adjusting "original" series stored in m$decomposition$components[, 'y_cmp' ] has different values for some periods than the original one. I used tramoseats function with following parameters:
Could you please help mi to fix my issue?