Github user edi-bice commented on a diff in the pull request:
https://github.com/apache/incubator-samoa/pull/48#discussion_r58394751
--- Diff: samoa-api/src/main/java/org/apache/samoa/streams/FileStream.java
---
@@ -52,9 +49,18 @@
's', "Source Type (HDFS, local FS)", FileStreamSource.class,
"LocalFileStreamSource");
+ public IntOption classIndexOption = new IntOption("classIndex", 'c',
+ "Class index of data. 0 for none or -1 for last attribute in
file.", -1, -1, Integer.MAX_VALUE);
+
+ private FloatOption floatOption = new FloatOption("classWeight", 'w',
"", 1.0);
--- End diff --
Yes, it would. The class weight option is indeed implemented via instance
weights.
Most machine learning algorithms focus on total error and in extremely
imbalanced scenarios (fraud, terrorism, disease) would fail to detect the
sparse class which is really what we're after. Class weighting allows one to
incorporate apriori knowledge of the imbalance. For example sklearn, R e1071
SVM packages have class weights options.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
