1
$\begingroup$

Can I decompose a time series with multiple seasonalities (an msts object) using tbats (in the forecast package for R) and get the random component of it? Just as getting the random component using the decompose function for ts objects?

Improve this question
$\endgroup$

1 Answer 1

Reset to default
1
$\begingroup$

You can use the function tbats.components on a tbats model fit. It will yield a matrix of time series containing the level, the slope and the season of the fitted model, as well as the actual observations. You can then simply subtract the components from the observations to obtain the residuals:

> library(forecast)
> fit <- tbats(USAccDeaths)

> foo <- tbats.components(fit)

> head(foo)
     observed    level     slope     season
[1,]     9007 9538.066 -43.81329  -823.6141
[2,]     8106 9590.489 -47.90006 -1548.1921
[3,]     8928 9642.926 -52.03042  -781.2602
[4,]     9137 9636.436 -52.12401  -528.4124
[5,]    10017 9661.657 -54.41414   304.7762
[6,]    10826 9848.610 -67.92587   815.5472

> foo[,"observed"]-foo[,"level"]-foo[,"slope"]-foo[,"season"]
              Jan          Feb          Mar          Apr          May
1973  336.3616979  111.6033706  118.3650491   81.1007939  104.9810786
1974 -222.9725629 -120.0746460   65.3314214   84.8823480 -199.9296274
1975   88.0935771  -12.2533927   20.1954758 -210.2334112  194.5652535
1976   91.6890807  225.6898321 -108.3721742 -104.6956780 -128.2708712
1977   18.9462879  -59.0440990  -48.6638245    9.8410292  -38.2662635
1978  -43.2497726 -132.1089133  -25.8376148   29.8293887   34.8255629
              Jun          Jul          Aug          Sep          Oct
1973  229.7689357    9.7833829   97.6213834  114.7690243   61.5528845
1974   23.6012054 -100.8973484   99.0247537   33.6304182   37.2522537
1975  -59.5136479 -162.8834589   -5.6052057 -131.8664155 -149.6577923
1976 -168.0888948   16.7956107 -130.6105420 -127.4818202  -48.5429559
1977  -81.8158991  155.2481713 -232.3853212  -95.3328947   18.0666666
1978  -82.3539729   38.3254629    8.2383295  149.0925592  -93.5561732
              Nov          Dec
1973  -24.4337637 -201.9057433
1974   70.8572386 -105.5657300
1975    5.6164054 -211.3018369
1976  -94.9396877  127.1735861
1977  -34.9677557   65.1114921
1978    0.1845764  136.0468109
Improve this answer
$\endgroup$
2
  • $\begingroup$ Thank you very much sir. I want to guess the number of AR and MA terms of this time series based on the ACF and PACF of the residual series. But the x axis of those plots ranges from 0.000 - 0.005. Please advice me how to initially guess the AR and MA terms of a time series with multiple seasonalities $\endgroup$
    – Deshani
    May 24, 2016 at 8:39
  • $\begingroup$ TBATS is not an ARIMA model - it's a type of exponential smoothing. $\endgroup$
    – Stephan Kolassa
    May 24, 2016 at 9:42

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.