Hello,
I have interest in weight my samples for linear learners such as
LinearRegression, LassoLarsCV, and friends.
I found an example for SVM based on per-sample C (
http://scikit-learn.org/stable/auto_examples/svm/plot_weighted_samples.html),
but this isn't the kind of learner I want to use. And sample_weight doesn't
seem to work on LassoLarsCV or other related linear methods I've tested it
on.
I tried pre-weighting based on weighted LS math (
http://en.wikipedia.org/wiki/Least_squares#Weighted_least_squares), and it
sort of worked in simple tests (where I either scaled or repeated samples),
but it doesn't seem to be working in my actual use case.
I also found a recent mailing list thread on recommendations for logistic
regression, recommending perhaps a particular PR branch (
https://github.com/scikit-learn/scikit-learn/pull/2784), but that still
seems to be going down the SVM route.
Is there any support in sklearn for weighted samples in simple linear
regression that I've missed? Or should my pre-weighting work after all (and
therefore I'm likely just doing it wrong)?
Thanks,
Tom Palmer
------------------------------------------------------------------------------
Android apps run on BlackBerry 10
Introducing the new BlackBerry 10.2.1 Runtime for Android apps.
Now with support for Jelly Bean, Bluetooth, Mapview and more.
Get your Android app in front of a whole new audience. Start now.
http://pubads.g.doubleclick.net/gampad/clk?id=124407151&iu=/4140/ostg.clktrk
_______________________________________________
Scikit-learn-general mailing list
Scikit-learn-general@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
