1. Title of Database: LED display domain
2. Sources:
(a) Breiman,L., Friedman,J.H., Olshen,R.A., & Stone,C.J. (1984).
Classification and Regression Trees. Wadsworth International
Group: Belmont, California. (see pages 43-49).
(b) Donor: David Aha
(c) Date: 11/10/1988
3. Past Usage: (many)
1. CART book (above):
-- Optimal Bayes classification rate: 74%
-- CART decision tree algorithm: 71% (resubstitution estimate)
-- Nearest Neighbor Algorithm: 71%
-- 200 training and 5000 test instances
2. Quinlan,J.R. (1987). Simplifying Decision Trees. In International
Journal of Man-Machine Studies (to appear).
-- C4 decision tree algorithm: 72.6% (using pessimistic pruning)
-- 2000 training and 500 test instances
3. Tan,M. & Eshelman,L. (1988). Using Weighted Networks to Represent
Classification Knowledge in Noisy Domains. In Proceedings of the
5th International Conference on Machine Learning, 121-134, Ann
Arbor, Michigan: Morgan Kaufmann.
-- IWN system: 73.3% (using the And-OR classification algorithm)
-- 400 training and 500 test cases
4. Relevant Information Paragraph:
This simple domain contains 7 Boolean attributes and 10 concepts,
the set of decimal digits. Recall that LED displays contain 7
light-emitting diodes -- hence the reason for 7 attributes. The
problem would be easy if not for the introduction of noise. In
this case, each attribute value has the 10% probability of having
its value inverted.
It's valuable to know the optimal Bayes rate for these databases.
In this case, the misclassification rate is 26% (74% classification
accuracy).
5. Number of Instances: chosen by the user.
6. Number of Attributes: 7 (all Boolean-valued)
7. Attribute Information:
-- All attribute values are either 0 or 1, according to whether
the corresponding light is on or not for the decimal digit.
-- Each attribute (excluding the class attribute, which is an
integer ranging between 0 and 9 inclusive) has a 10% percent
chance of being inverted.
8. Missing Attribute Values: None
9. Class Distribution: 10% (Theoretical)
-- Each concept (digit) has the same theoretical probability
distribution. The program randomly selects the attribute.