Combined 5x2cv $F$-Test for Comparing Supervised Classification Learning Algorithms
| Type of publication: | Idiap-RR |
| Citation: | Alpa98 |
| Number: | Idiap-RR-04-1998 |
| Year: | 1998 |
| Institution: | IDIAP |
| Note: | Submitted for publication |
| Abstract: | Dietterich (1998) reviews five statistical tests proposing the 5x2cv t test for determining whether there is a significant difference between the error rates of two classifiers. In our experiments, we noticed that the 5x2cv t test result may vary depending on factors that should not affect the test and we propose a variant, the combined 5x2cv F test, that combines multiple statistics to get a more robust test. Simulation results show that this combined version of the test has lower Type I error and higher power than 5x2cv proper. |
| Userfields: | ipdmembership={learning}, |
| Keywords: | |
| Projects: |
Idiap |
| Authors: | |
| Added by: | [UNK] |
| Total mark: | 0 |
|
Attachments
|
|
|
|
|
Notes
|
|
|
|
|
