Hi Guys,
Does any one have the experience on how to normalise(or we say standardise)
the periodic data, for example, the attached data.
I am not going to use it for forecasting, rather than treat it as an input
variable for another model with the other time series data.
e.g. model the volatility of return series, r(t), one can add trade volume
as an exogenous variable in the conditional variance equation. IF the volume
data has the characteristic of periodicity, should it be normalised?
I know the moving average is a method to detect the seasonality or periodic,
however, it will reduce the data observations, e.g. we have one series {a1,
a2, ..... , a99}, after taking the moving average (3 point) it could be only
one third of the {a1', a2',...,a33'} series.
Appreciate for any comment.
Cheers
Mam
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://stat.ethz.ch/pipermail/r-sig-finance/attachments/20081024/40499bd7/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: periodic data.xls
Type: application/vnd.ms-excel
Size: 27648 bytes
Desc: not available
URL: <https://stat.ethz.ch/pipermail/r-sig-finance/attachments/20081024/40499bd7/attachment.xls>
about periodicity data
2 messages · Michael Sun, Brian G. Peterson
Michael Sun wrote:
Hi Guys, Does any one have the experience on how to normalise(or we say standardise)the periodic data, for example, the attached data.
This looks a lot like seasonality. Perhaps that could aid your searching, as much has been written, including on this list, on dealing with Seasonality of data in R. Without looking, I recall that Shumway and Stoffer cover this topic. R packages uroot, partsm, and ast may prove useful. Also the base 'stats' package contains the function decompose() and a few others for dealing with seasonality. Please report back so other can benefit from whatever you figure out. Regards, - Brian