Letter Recognition
The objective is to identify each of a large number of black-and-white rectangular pixel displays as one of the 26 capital letters in the English alphabet. The character images were based on 20 different fonts and each letter within these 20 fonts was randomly distorted to produce a file of 20.000 unique stimuli. Each stimulus was converted into 16 primitive numerical attributes (statistical moments and edge counts) which were then scaled to fit into a range of integer values from 0 through 15.
Data Set Information
| Data Set Characteristics | Multivariate |
| Attribute Characteristics | Integer |
| Number of Attributes | 16 |
| Number of Instances | 20.000 |
| Associated Tasks | Classification |
Results
We are going to measure the accuracy rate into the test subset(4.000 instances)
| Technique | Test Rate |
|---|---|
| LDA | 0.8955 |
| QDA | 0.9497 |
| KNN | 0.9641 |
| Tree(simple) | 0.4799 |
| Bagging | 0.9454 |
| Random Forests | 0.9915 |
| Boosting | 0.5805 |
| SVM | 0.9487 |
Source
David J. Slate
Odesta Corporation;
1890 Maple Ave; Suite 115;
Evanston, IL 60201