5 messages · Gabor Grothendieck, elisia, Girish A.R. +1 more
how can I eliminate the influence of the festivities in a time series with daily data?I tried to remove them and replace their value with a value of interpolation using na.approx (). There is an alternative method?
View this message in context: http://www.nabble.com/holidays-effect-tp21830785p21830785.html Sent from the R help mailing list archive at Nabble.com.
One possibility if you don't have to have days is to reduce it to a weekly or monthly series. On Wed, Feb 4, 2009 at 8:46 AM, elisia
how can I eliminate the influence of the festivities in a time series with daily data?I tried to remove them and replace their value with a value of interpolation using na.approx (). There is an alternative method? -- View this message in context: http://www.nabble.com/holidays-effect-tp21830785p21830785.html Sent from the R help mailing list archive at Nabble.com.
______________________________________________ R-help at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
Unfortunately, my aim is to identify outliers in a time series, and I would like the holidays, which have in general a higher value and shall not affect such research. If I run the detection of outliers on the whole time series, while maintaining the real values of the holidays, I get results just as the values corresponding to the festivities. How can I solve this problem, by maintaining the daily data?
One possibility if you don't have to have days is to reduce it to a weekly or monthly series. On Wed, Feb 4, 2009 at 8:46 AM, elisia <elisabetta.fabris at guest.telecomitalia.it> wrote:
how can I eliminate the influence of the festivities in a time series with daily data?I tried to remove them and replace their value with a value of interpolation using na.approx (). There is an alternative method? -- View this message in context: http://www.nabble.com/holidays-effect-tp21830785p21830785.html Sent from the R help mailing list archive at Nabble.com.
______________________________________________ R-help at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
______________________________________________ R-help at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
View this message in context: http://www.nabble.com/holidays-effect-tp21830785p21831641.html Sent from the R help mailing list archive at Nabble.com.
Just an extension of the query posed by the OP --- Similar problem arises in the case when one has to deal with weekly data spanning 2 or more years, and one of the years happens to have 53 weeks because it is a leap year (2004, for ex.). In a sci.stats newsgroup where I had posed this problem for recommendations, one suggestion was to check out time series calendar adjustment topics. I know some commercial software have the option of trading day adjustment where the adjustment is obtained by regression on the days of the week for the months under consideration. I'm currently exploring R time series packages. Any thoughts on this would be appreciated. Thanks! -Girish
One possibility if you don't have to have days is to reduce it to a weekly or monthly series. On Wed, Feb 4, 2009 at 8:46 AM, elisia <elisabetta.fab... at guest.telecomitalia.it> wrote:
how can I eliminate the influence of the festivities in a time series with daily data?I tried to remove them and replace their value with a value of interpolation using na.approx (). There is an alternative method? -- View this message in context:http://www.nabble.com/holidays-effect-tp21830785p21830785.html Sent from the R help mailing list archive at Nabble.com.
Gabor Grothendieck ha scritto:
One possibility if you don't have to have days is to reduce it to a weekly or monthly series.
Alternatively you can put a dummy variable (1=holiday and zero otherwise) in the regression model for your response. For instance, you could use the xreg argument of the arima() function. This allows to avoid aggregation of your data which, in general, is not recommended.. best, vito
On Wed, Feb 4, 2009 at 8:46 AM, elisia <elisabetta.fabris at guest.telecomitalia.it> wrote:
how can I eliminate the influence of the festivities in a time series with daily data?I tried to remove them and replace their value with a value of interpolation using na.approx (). There is an alternative method? -- View this message in context: http://www.nabble.com/holidays-effect-tp21830785p21830785.html Sent from the R help mailing list archive at Nabble.com.
______________________________________________ R-help at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
______________________________________________ R-help at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
==================================== Vito M.R. Muggeo Dip.to Sc Statist e Matem `Vianelli' Universit? di Palermo viale delle Scienze, edificio 13 90128 Palermo - ITALY tel: 091 6626240 fax: 091 485726/485612 http://dssm.unipa.it/vmuggeo