Deprecate PARAM_CUTOFF parameters in Default*DataWriterFactory classes

[GoogleCodeExporter]

With Lee's implementation of feature transformations, I think we should really 
deprecate all the PARAM_CUTOFF parameters in the Default*DataWriterFactory 
classes. All these do is trim the feature lookup table so that at 
classification time, some features can't be used - however, at training time, 
all the features are still included. In the worst case scenario, the model 
would learn to put weight only on the features that aren't trimmed, and then 
the classifier would classify everything as 0.0!

So I think these are a serious design mistake. So serious that maybe we should 
skip deprecation and just rip them out entirely.

Original issue reported on code.google.com by steven.b...@gmail.com on 28 Mar 2012 at 11:07

[GoogleCodeExporter]

These were essentially fixed alongside Issue 281 - all the DataWriterFactory 
classes were deprecated, including the ones that contain a PARAM_CUTOFF.

Original comment by steven.b...@gmail.com on 25 Apr 2012 at 2:19

• Changed state: Fixed
• Added labels: ****
• Removed labels: ****

[GoogleCodeExporter]

Original comment by steven.b...@gmail.com on 5 Aug 2012 at 8:49

• Added labels: Component-ml, Milestone-1.2
• Removed labels: ****