I am trying to model a multi variate time series model. (i.e. Multivariate ARIMA). The general format of the model is in the shape of a budget model like:
W(t) = f [ X(t), Y(t), Z(t), T(t) ]
were W(t) shows the jump and drop that I mentioned. There is jump and a following drop in the historical dependent variable (i.e. Output). But this record is not just on one data and records seem to persist on this behavior for a medium period (i.e. several years after jump and then a following drop). I have attached the diagram of the time series to this post. As soon as the independent variables spouse to cause or track down the jump and drop; and since I am going to make them independent from each other. Do I have to remove the jump from the data or not?
I think changes in input variables should track or result in that jump and drop and it is not proper to remove the jump and the drop.
Dear Babak
you dont have to remove the jumps and drops from the data as your model will automatically incorporate these changes into the model system parameters. These features may be low frequency trends that may be important clues in your data you dont want to lose. We often have these in computer simulation models for solar thermal and solar power systems in smart microgrid configurations.
Regards
Gerro
Gerro, thanks for your comment.
actually I am hearing many advices from different people to traditionally remove the trend from data. But just like you said, it doesn't make sence and I think it is an important information related to data.
Actually it is a budget model and it is a closed one. The final shape of the model would be like:
Z(t) = aX(t)+ bY(t) - cW(t) + dS(t)
therefore, as you said X,Y,W and S should be able to model any changes in Z and saving any changes in dependent variable is important.
Hi,
To remove jump and drop in a Multivariate time series model:
Kindly refer the following links:
1. https://stat.ethz.ch/education/semesters/ss2014/atsa/Scriptum_v140523.pdf
2. http://www.stata.com/manuals13/ts.pdf
3. https://cran.r-project.org/web/packages/timeSeries/timeSeries.pdf
Thank you Senthilvel.
But do you really think that it is necessary to remove such an information from data?
Don't remove the jumps. But you can explain it. Every information or data is important.
I agree with you Seuk, can you give me some referrences?
I realy need a concrete referrence for this manipulution. I.e. Not removing jump and drop
Before removing any data, think carefully about the data quality. Is there any reason for these changes to be associated with some sort of change in the measuring, siting, storage, retrieval, or QC processes?
Geoffrey.
It is more like an extreme event, rather than precision error in data records. I understand what you mean but I am almost sure that there is not a significant error in recorded data.
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis