Limitations of Rule Induction

  1. ''Flat`` classification rules produced with no justification facility
  2. Lots of training examples are necessary
  3. Training must be noise-free
  4. Each training example must be described using all the features
  5. Classifications and features are static sets
  6. Rules produced do not distinguish between correlation and causality

Contents    Page-10    Prev    Next    Page+10    Index