Hi,
I've been playing with the SGD classifier (using the current master
branch), and found that when there is a large range in x, the classifier
fails. For example, when running the example script found here:
http://scikit-learn.org/0.10/auto_examples/linear_model/plot_sgd_ols.html
If I replace the line:
xmin, xmax = -5, 5
with:
xmin, xmax = 0, 100
then the regression fails (full traceback copied below). If the range
is set to (0, 50), the call does not produce an error, but leads to a
line which clearly does not fit the data (intercept on order -2E30).
Any ideas about what is going on here?
Jake
"/usr/local/lib/python2.6/dist-packages/scikit_learn-0.12_git-py2.6-linux-i686.egg/sklearn/linear_model/stochastic_gradient.py",
line 737, in fit
coef_init, intercept_init)
File
"/usr/local/lib/python2.6/dist-packages/scikit_learn-0.12_git-py2.6-linux-i686.egg/sklearn/linear_model/stochastic_gradient.py",
line 672, in _partial_fit
self._fit_regressor(X, y, sample_weight, n_iter)
File
"/usr/local/lib/python2.6/dist-packages/scikit_learn-0.12_git-py2.6-linux-i686.egg/sklearn/linear_model/stochastic_gradient.py",
line 786, in _fit_regressor
intercept_decay)
File "sgd_fast.pyx", line 423, in
sklearn.linear_model.sgd_fast.plain_sgd
(sklearn/linear_model/sgd_fast.c:6713)
ValueError: floating-point under-/overflow occured.
------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Scikit-learn-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
