scikit-sos
scikit-sos copied to clipboard
jeroenjanssens
Divide By zero
/anaconda3/lib/python3.6/site-packages/scikit_sos-0.1.10-py3.6.egg/sksos/sos.py:132: RuntimeWarning: overflow encountered in multiply A = np.exp(-D * beta) /anaconda3/lib/python3.6/site-packages/scikit_sos-0.1.10-py3.6.egg/sksos/sos.py:88: RuntimeWarning: overflow encountered in multiply beta[i] = beta[i] * 2.0 /anaconda3/lib/python3.6/site-packages/scikit_sos-0.1.10-py3.6.egg/sksos/sos.py:132: RuntimeWarning: invalid value encountered in multiply A = np.exp(-D * beta) /anaconda3/lib/python3.6/site-packages/scikit_sos-0.1.10-py3.6.egg/sksos/sos.py:134: RuntimeWarning: divide by zero encountered in log H = np.log(sumA) + beta * np.sum(D * A) / sumA /anaconda3/lib/python3.6/site-packages/scikit_sos-0.1.10-py3.6.egg/sksos/sos.py:134: RuntimeWarning: invalid value encountered in true_divide H = np.log(sumA) + beta * np.sum(D * A) / sumA
how did you resolve it @nakulstar95 ?
I'm getting the following error,
C:\Anaconda\lib\site-packages\sksos\sos.py:132: RuntimeWarning:
overflow encountered in multiply
C:\Anaconda\lib\site-packages\sksos\sos.py:88: RuntimeWarning:
overflow encountered in multiply
C:\Anaconda\lib\site-packages\sksos\sos.py:132: RuntimeWarning:
invalid value encountered in multiply
@nakulstar95 @sskarkhanis any suggestions on fixing this issue? I am running into this issue as well when all my values are 1.0. thanks!
@nakulstar95 @sskarkhanis any suggestions on fixing this issue? I am running into this issue as well when all my values are 1.0. thanks!
I've moved on to using the PyOD library for using outlier detection methods.
https://github.com/yzhao062/pyod
It does allow you to use SOS but in my experience on using it even with pyOD, SOS doesnt work well and can be buggy. Good news is that you can use the other methods there which work.