Name | Modified | Size | Downloads / Week |
---|---|---|---|
Parent folder | |||
README | 2012-06-30 | 350 Bytes | |
CfsSubset.java | 2012-06-30 | 8.0 kB | |
CorrelationSubset.java | 2012-06-30 | 5.1 kB | |
NonemptySubset.java | 2012-06-30 | 2.9 kB | |
FeatureSelection.java | 2012-06-30 | 2.4 kB | |
Totals: 5 Items | 18.8 kB | 0 |
The use of CfsSubset should be straightforward. However, in case of too many features (over 3000-5000), it may become slow even if it is well optimized. In that case the more basic selection algorithms should be used to reduce number of features. The NonemptySubset algorithm is used to clean up features with a single value for all records.