What is more efficient, autocorrelation or SSA?

  • Context: Graduate 
  • Thread starter Thread starter Adel Makram
  • Start date Start date
  • Tags Tags
    Autocorrelation
Click For Summary

Discussion Overview

The discussion revolves around the efficiency of two methods for extracting patterns in time series analysis: autocorrelation and singular spectrum analysis (SSA). Participants explore the definition of "efficiency" in this context, considering factors such as real-time analysis and the completeness of information extracted from the data.

Discussion Character

  • Debate/contested

Main Points Raised

  • Some participants question the definition of "efficient," suggesting that clarity on what the method needs to achieve and under what constraints is essential.
  • One participant proposes that the fastest method might simply involve visually inspecting the data and making guesses.
  • Another participant emphasizes that by "efficient," they refer to the algorithm's ability to extract all possible information about the spectral components of the time series, while also noting the importance of real-time analysis.
  • It is suggested that neither autocorrelation nor SSA will yield "all possible" information, as both are approximate methods.
  • Participants note that the speed of each method may depend on the specific target and the nature of the data, and that a quantitative assessment could be made by calculating the number of machine cycles required for each method.
  • One participant expresses skepticism that either method is intrinsically more efficient than the other.

Areas of Agreement / Disagreement

Participants generally do not agree on a definitive answer regarding which method is more efficient, and multiple competing views remain about the definition of efficiency and the capabilities of each method.

Contextual Notes

Limitations in the discussion include the lack of consensus on the definition of "efficiency," the dependence on specific constraints, and the acknowledgment that both methods are approximate and may not capture all information.

Adel Makram
Messages
632
Reaction score
15
What is more efficient in extracting the pattern in a time series analysis, autocorrelation or singular spectrum analysis?
 
Physics news on Phys.org
Define "efficient" - then you will have your answer.
What does the method have to achieve and under what constraints?

i.e. the fastest (quickest real time) method is to look at the data and guess.
 
Simon Bridge said:
Define "efficient" - then you will have your answer.
What does the method have to achieve and under what constraints?

i.e. the fastest (quickest real time) method is to look at the data and guess.
By efficient, I meant the ability of the algorithm to get all possible information about the spectral components of the time series. Real time analysis is an important concern too.
 
Well I doubt either approach will get "all possible" information... they are both approximate methods.
Which is faster in real time will depend on the target and the data. But here you have the ability to get a quantitative assessment... i.e. by working out the number of machine cycles needed for each method.

Notice, I am not answering your question so much as trying to get you to think about it more clearly so you can answer it yourself.

There is a reason there is more than one way to skin this particular cat.
I suspect the short answer is that neither is intrinsically more efficient than the other.
 

Similar threads

  • · Replies 7 ·
Replies
7
Views
2K
  • · Replies 1 ·
Replies
1
Views
2K
  • · Replies 1 ·
Replies
1
Views
7K
  • · Replies 1 ·
Replies
1
Views
1K
  • · Replies 2 ·
Replies
2
Views
3K
  • · Replies 0 ·
Replies
0
Views
3K
  • · Replies 41 ·
2
Replies
41
Views
8K
  • · Replies 14 ·
Replies
14
Views
2K
  • · Replies 6 ·
Replies
6
Views
2K
  • · Replies 1 ·
Replies
1
Views
2K