2

I'm working on a project that involves concept drift detection for a time series. Are there any well-known techniques/methods/algorithms that are known to be effective for this sort of problem?

Currently, I was thinking of using a Kalman filter approach.

DaL
  • 4,462
  • 3
  • 16
  • 27
Glassjawed
  • 457
  • 3
  • 13

1 Answers1

1

You might find these useful:

  • Learn++.NSE from Prof.Robi Polikar. This is his official page which offers lots of additional information, papers on the topic etc. and it is a great starting point.
  • The GitHub repository of gditzler contains an implementation of incremental learners in MATLAB.

Additionally the Concept Drift wiki page, under sections Projects and Meetings has lots of other materials.hth.

UPDATE
I came across a paper

  • "Incremental Learning of Support Vector Machines" by Classiffier Combining by Wen and Lu.

stating superior performance and efficiency than concept drift methods.

davidski
  • 191
  • 5
  • 1
    whenever possible please provide enough information to still be able to access the information in case the link dies (e.g., instead of "this guy", provide the user name) – Antoine Mar 29 '16 at 14:24
  • yes, you are right. fixed. – davidski Mar 29 '16 at 15:08