2

After fitting my time series with an ARIMA model, I want to test outliers in the residuals' series. Are there any functions in R that could do this test and furtherly test whether the outlier is additive or innovational, seasonal or just one pulse?

Nick Stauner
  • 11,558
  • 5
  • 47
  • 105
user3060702
  • 61
  • 6
  • I dont think there is one. There is detctIO which detects Innovational Outlier in TSA package. See also posts http://stats.stackexchange.com/questions/62237/how-do-i-incorporate-an-innovative-outlier-at-observation-48-in-my-arima-model and http://stats.stackexchange.com/questions/28003/how-to-specify-pulses-level-shifts-in-data-when-creating-arima-in-r – forecaster Jan 06 '14 at 19:28
  • Check out this [question](http://stats.stackexchange.com/questions/1142/simple-algorithm-for-online-outlier-detection-of-a-generic-time-series) as well, specially the R code by prof. Rob Hyndman. – Stat Jan 07 '14 at 12:51
  • Also, commercially available packages such as SPSS, SAS, and autobox does this automatically for you. http://support.sas.com/documentation/cdl/en/etsug/60372/HTML/default/viewer.htm#etsug_arima_sect045.htm http://www.autobox.com/cms/index.php/afs-university/autobox-examples/modeling-with-autobox#_Toc509396187 http://pic.dhe.ibm.com/infocenter/spssstat/v21r0m0/index.jsp?topic=%2Fcom.ibm.spss.statistics.help%2Falg_tsmodel_outliers.htm – forecaster Jan 08 '14 at 02:18

1 Answers1

1

TSA package features detectIO and detectAO, but (although you haven't stated what you're trying to do, just FYI) the arimax function will only allow you to fit a model, not forecast with it.

Robert Hyndman's code in the question linked to by Stat does not identify 'types' of outliers or possible dynamic impact.

krcooke
  • 143
  • 1
  • 1
  • 9