Appropriate Problems for Decision Tree Learning - Artificial intelligence
It is a expert job in AI to select accurately the right learning representation for a particular learning job. As convoluted by Tom Mitchell, decision tree learning is best suited to problems with these characteristics:
- The background concepts explain the examples in terms of attribute-value pairs, and the values for every feature range over finitely various fixed promises.
- The idea to be learned (Mitchell calls it the target function) has separate values.
- Disjunctive descriptions should be required in the answer.
In addition to that, decision tree learning is robust to mistakes in the data. In particular, it will act well in the light of (i) mistakes in the categorization occurrences given (ii) errors in the features-value pairs provided and (iii) missing values for fix features for fix examples.