Commit Inicial

data/tanveer/led-display/led-creator.names

@@ -0,0 +1,55 @@
+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.