test failure test_arima.test_small_data #1415

Closed
josef-pkt opened this Issue Feb 21, 2014 · 3 comments

Projects

None yet

2 participants

@josef-pkt
Member

test failure in test_arima.test_small_data with SVD did not converge
https://launchpadlibrarian.net/167128211/buildlog_ubuntu-saucy-amd64.statsmodels_0.6.0~ppa18~revno-1556~ubuntu13.10.1_UPLOADING.txt.gz

this is only amd64, there is no failure on i386

for previous problems with test_small_data see #1149

======================================================================
ERROR: statsmodels.tsa.tests.test_arima.test_small_data
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/usr/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
    self.test(*self.arg)
  File "/build/buildd/statsmodels-0.6.0~ppa18~revno/debian/python-statsmodels/usr/lib/python2.7/dist-packages/statsmodels/tsa/tests/test_arima.py", line 1894, in test_small_data
    res = mod.fit(disp=0, start_params=[.1, .1, .1, .1])
  File "/build/buildd/statsmodels-0.6.0~ppa18~revno/debian/python-statsmodels/usr/lib/python2.7/dist-packages/statsmodels/tsa/arima_model.py", line 885, in fit
    callback=callback, **kwargs)
  File "/build/buildd/statsmodels-0.6.0~ppa18~revno/debian/python-statsmodels/usr/lib/python2.7/dist-packages/statsmodels/base/model.py", line 385, in fit
    hess=hess)
  File "/build/buildd/statsmodels-0.6.0~ppa18~revno/debian/python-statsmodels/usr/lib/python2.7/dist-packages/statsmodels/base/model.py", line 571, in _fit_mle_lbfgs
    **extra_kwargs)
  File "/usr/lib/python2.7/dist-packages/scipy/optimize/lbfgsb.py", line 185, in fmin_l_bfgs_b
    **opts)
  File "/usr/lib/python2.7/dist-packages/scipy/optimize/lbfgsb.py", line 312, in _minimize_lbfgsb
    f, g = func_and_grad(x)
  File "/usr/lib/python2.7/dist-packages/scipy/optimize/lbfgsb.py", line 257, in func_and_grad
    f = fun(x, *args)
  File "/build/buildd/statsmodels-0.6.0~ppa18~revno/debian/python-statsmodels/usr/lib/python2.7/dist-packages/statsmodels/base/model.py", line 356, in <lambda>
    f = lambda params, *args: -self.loglike(params, *args) / nobs
  File "/build/buildd/statsmodels-0.6.0~ppa18~revno/debian/python-statsmodels/usr/lib/python2.7/dist-packages/statsmodels/tsa/arima_model.py", line 695, in loglike
    return self.loglike_kalman(params, set_sigma2)
  File "/build/buildd/statsmodels-0.6.0~ppa18~revno/debian/python-statsmodels/usr/lib/python2.7/dist-packages/statsmodels/tsa/arima_model.py", line 705, in loglike_kalman
    return KalmanFilter.loglike(params, self, set_sigma2)
  File "/build/buildd/statsmodels-0.6.0~ppa18~revno/debian/python-statsmodels/usr/lib/python2.7/dist-packages/statsmodels/tsa/kalmanf/kalmanfilter.py", line 646, in loglike
    R_mat, T_mat)
  File "kalman_loglike.pyx", line 119, in statsmodels.tsa.kalmanf.kalman_loglike.kalman_loglike_double (statsmodels/tsa/kalmanf/kalman_loglike.c:3684)
  File "kalman_loglike.pyx", line 38, in statsmodels.tsa.kalmanf.kalman_loglike.kalman_filter_double (statsmodels/tsa/kalmanf/kalman_loglike.c:1810)
  File "/usr/lib/python2.7/dist-packages/numpy/linalg/linalg.py", line 1574, in pinv
    u, s, vt = svd(a, 0)
  File "/usr/lib/python2.7/dist-packages/numpy/linalg/linalg.py", line 1323, in svd
    raise LinAlgError('SVD did not converge')
LinAlgError: SVD did not converge
@jseabold
Member

FFS. This an example of the unhelpful SVD message that crops up often in ARIMA. I don't know how to debug this without being able to replicate. Possibly related to ubuntu shipping with numpy built against openblas now (as per the ML conversation)? Total shot in the dark.

@josef-pkt
Member

that machine uses
numpy 1.7.1
libblas3 amd64 1.2.20110419-5
liblapack3 amd64 3.4.2+dfsg-2

I have no idea what that means.

@jseabold jseabold added this to the 0.5.1 milestone Apr 8, 2014
@josef-pkt
Member

#1697 adds a better starting value (just for the constant) to fix problems for Snake Charmer, issue #1689

closing this, let's hope it's the last time we see it

@josef-pkt josef-pkt closed this Jun 18, 2014
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment