3

In the context of unevenly spaced (multiple) time series (USTS), are there any classical approaches? If they were evenly spaced, we would try ARIMA, or VAR, or even State-Space models.

I've been searching for bibliography, and from what I could tell, it seems that this field of USTS hasn't been as well developed as the evenly spaced time series, and it's a bit hard to tell which references are the 'classical' ones in this field.

Also, I'm not necessarily restricting the question to those models more similar to classical one like ARIMA, etc. I'm also interested in Machine Learning (and DL) models that are capable of dealing with USTS.

P.S.: There's this highly voted question on Cross Validated, but it was asked almost 10 years ago. Maybe things have improved since then...

An old man in the sea.
  • 5,640
  • I read (IIRC it was on Wikipedia) that there isn't much development in the field of irregular time series, considering our advances in computing power. However, there's a classical way of analysing unevenly spaced time series and that is the frequency domain approach, for example spectrum analysis, which help us exploit the traits and whatnot for each frequency (i.e., different intervals). But that might not give you information that you'd expect from a time domain approach. – stucash May 07 '22 at 18:32

0 Answers0