Hi Jane.
What is exactly you want to do? If you want to smooth a time series, you have different ways to do it. I usually use or rumming average filtering or a butterwoth filter.
Let me know which is your question and I will try to help you if I know how.
Àngel.
Using discrete time filters for comparing all these objects, all these forms could write as a discrete time filter, iir, fir, or design specific filters, and after you compare with standar error metrics, or kolmogorov-smirnov tests and cross correlations.
Note: i used this way for comparing, armax, garch, kalman filters and exponential forms, and others things
Apologies for responding so late in your time line just thought it worth logging a view here. Hope it helps a little... Sometimes it is valuable to ask what is the objective of your work and then to look very seriously at the implied assumptions of your possible (proposed) Method[s]. No doubt there is much written on this but Scott Armstrong and Fed Callopy in Vol 12 of Journal of Forecasting talk about some important things around current trend vs historical trend and if i recall correctly on the use methods based on this understanding.
best regards Murray MacRae
- Badges
- Science topic
- Similar topics
- Statistics
- Time Series