PS: I started this
thread<http://stackoverflow.com/questions/18213789/feature-selection-with-linearsvc>in
StackOverflow with the hope to clarify the original question.
Josh
On Tue, Aug 13, 2013 at 11:24 AM, Josh Wasserstein
<[email protected]>wrote:
> When running feature selection with LinearSVC for some data that I have
> X_new = LinearSVC(C=0.01, penalty="l1", dual=False).fit_transform(X, y)
>
> I get:
> Invalid threshold: all features are discarded.
>
> Even though the rfecv and anova methods work fine.
>
> According to the documentation:
> "Threshold takes the value None by default. If None and if available, the
> object attribute threshold is used. Otherwise, “mean” is used by default"
>
> Which object is the documentation referring to in the context of my
> example above? And why do I get invalid threshold in my case?
>
> Finally, is there a way to know which specific features (columns) were
> discarded? Does the method give a score per feature? or does it just make
> binary decisions on which feature are filtered in vs out?
>
> Sorry I have so many questions about LinearSVC. I wonder if it would be
> useful to collect Q&A in some place.
>
> Thanks,
>
> Josh
>
>
>
------------------------------------------------------------------------------
Get 100% visibility into Java/.NET code with AppDynamics Lite!
It's a free troubleshooting tool designed for production.
Get down to code-level detail for bottlenecks, with <2% overhead.
Download for free and get started troubleshooting in minutes.
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
Scikit-learn-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
