Skip to content

Latest commit

 

History

History
executable file
·
16 lines (13 loc) · 1.25 KB

README.md

File metadata and controls

executable file
·
16 lines (13 loc) · 1.25 KB
Raw

Multi-label branch of MOA

Multi-label classification should work out-of-the-box in MOA. If you want to run some examples in a linux environment. There is a test.sh bash script on this branch. Follow these instructions:

  1. Either
    • Checkout the code from this repository (multilabel branch); or
    • Checkout the code from the main repository and download the following files:
      • meka-1.9.1-SNAPSHOT.jar (or compile it yourself from the Meka project
      • test.sh contains many examples
      • align.sh simply an auxilliary script for aligning output neatly on the terminal
  2. Check the CLASSPATH options in the beginning of tesh.sh
  3. Download a dataset, for example A-TMC7-REDU-X2-500.arff is the default one used in the script
  4. Run the script, e.g., ./test.sh 0 if you want to test the code from the 2012 paper Read, Bifet, Pfahringer, Holmes - Scalable and Efficient Multi-label Classification for Evolving Data Streams.

For more (older) information, and additional notes, see this link referenced in the paper