IF-THEN Rules for Classification

Represents knowledge of data in the form of IF-THEN rules.

These rules are more readable and easier to understand than large Decision Trees.

Problems

More than one rule is triggered.

• Size ordering → highest priority for toughest rule
• Class-based ordering →
• Rule-based ordering (decision list) → rules have to be applied in this particular order to avoid conflicts. Order can be established by some measure of quality or by experts

No rule is triggered.

• Fallback rule which will always be evaluated as the last rule