Compression-Based Pruning of Decision Lists

  title={Compression-Based Pruning of Decision Lists},
  author={Bernhard Pfahringer},
We deene a formula for estimating the coding costs of decision lists for propositional domains. This formula allows for multiple classes and both categorical and numerical attributes. For artiicial domains the formula performs quite satisfactory, whereas results are rather mixed and inconclusive for natural domains. Further experiments lead to a principled simpliication of the original formula which is robust in both artiicial and natural domains. Simple hill-climbing search for the most… CONTINUE READING

From This Paper

Figures, tables, and topics from this paper.
2 Citations
19 References
Similar Papers


Publications referenced by this paper.
Showing 1-10 of 19 references

Pruning Algorithms for Rule Learning, Osterreichisches Forschungsinstitut f ur Artiicial Intelligence

  • Pruning Algorithms for Rule Learning…
  • 1996

Practical Uses of the Minimum Description Length Principle in Inductive Learning

  • B Pfahringer
  • Practical Uses of the Minimum Description Length…
  • 1995

Similar Papers

Loading similar papers…