Bug#968210: statsmodels: arm* test failures/crashes

Sun Aug 16 18:22:27 BST 2020

The TestZeroInflatedModel_probit issue appears to be a failure to 
converge: as it does trigger warnings of that, and it's already skipped 
on i386, I intend to ignore it.

qemu-arm64:

 >>> a
<statsmodels.discrete.tests.test_count_model.TestZeroInflatedModel_probit 
object at 0x4000d31130>
 >>> a.res1.params
Traceback (most recent call last):
   File "<stdin>", line 1, in <module>
AttributeError: 'TestZeroInflatedModel_probit' object has no attribute 
'res1'
 >>> a.setup_class()
/usr/lib/python3/dist-packages/statsmodels/base/model.py:567: 
ConvergenceWarning: Maximum Likelihood optimization failed to converge. 
Check mle_retvals
   warn("Maximum Likelihood optimization failed to converge. "
 >>> a.res1.params
array([ 0.06225336, -0.64293239, -0.08217881,  0.00856726, -0.02679518,
         1.4823691 ])
 >>> 
alpha=np.ones(6);alpha[-2:]=0;res_reg=a.res1.model.fit_regularized(alpha=alpha*0.01,disp=0,maxiter=500)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:71: 
ConvergenceWarning: QC check did not pass for 4 out of 4 parameters
Try increasing solver accuracy or number of iterations, decreasing 
alpha, or switch solvers
   warnings.warn(message, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:144: 
ConvergenceWarning: Could not trim params automatically due to failed QC 
check. Trimming using trim_mode == 'size' will still work.
   warnings.warn(msg, ConvergenceWarning)
 >>> res_reg.params
array([ 0.        ,  0.        ,  0.        ,  0.        , -0.02679517,
         1.48236838])
 >>> 
alpha=np.ones(6);alpha[-2:]=0;res_reg=a.res1.model.fit_regularized(alpha=alpha*0.0,disp=0,maxiter=500)
 >>> res_reg.params
array([ 0.0622522 , -0.6429312 , -0.08217381,  0.00856762, -0.02679573,
         1.48237116])
 >>> 
alpha=np.ones(6);alpha[-2:]=0;res_reg=a.res1.model.fit_regularized(alpha=alpha*1e-6,disp=0,maxiter=500)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:71: 
ConvergenceWarning: QC check did not pass for 4 out of 4 parameters
Try increasing solver accuracy or number of iterations, decreasing 
alpha, or switch solvers
   warnings.warn(message, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:144: 
ConvergenceWarning: Could not trim params automatically due to failed QC 
check. Trimming using trim_mode == 'size' will still work.
   warnings.warn(msg, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:71: 
ConvergenceWarning: QC check did not pass for 4 out of 6 parameters
Try increasing solver accuracy or number of iterations, decreasing 
alpha, or switch solvers
   warnings.warn(message, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:144: 
ConvergenceWarning: Could not trim params automatically due to failed QC 
check. Trimming using trim_mode == 'size' will still work.
   warnings.warn(msg, ConvergenceWarning)
 >>> res_reg.params
array([ 0.06225361, -0.64293282, -0.08218008,  0.0085677 , -0.02679533,
         1.48236748])
 >>> 
alpha=np.ones(6);alpha[-2:]=0;res_reg=a.res1.model.fit_regularized(alpha=alpha*1,disp=0,maxiter=500)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:71: 
ConvergenceWarning: QC check did not pass for 2 out of 4 parameters
Try increasing solver accuracy or number of iterations, decreasing 
alpha, or switch solvers
   warnings.warn(message, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:144: 
ConvergenceWarning: Could not trim params automatically due to failed QC 
check. Trimming using trim_mode == 'size' will still work.
   warnings.warn(msg, ConvergenceWarning)
 >>> res_reg.params
array([ 0.        , -0.64270273, -0.08204933,  0.        , -0.02679321,
         1.48237335])
 >>> 
alpha=np.ones(6);alpha[-2:]=0;res_reg=a.res1.model.fit_regularized(alpha=alpha*100,disp=0,maxiter=500)
 >>> res_reg.params
array([ 0.05531257, -0.62025715, -0.06936752,  0.00781262, -0.02661252,
         1.48282813])
 >>> 
alpha=np.ones(6);alpha[-2:]=0;res_reg=a.res1.model.fit_regularized(alpha=alpha*1e-4,disp=0,maxiter=500)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:71: 
ConvergenceWarning: QC check did not pass for 4 out of 4 parameters
Try increasing solver accuracy or number of iterations, decreasing 
alpha, or switch solvers
   warnings.warn(message, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:144: 
ConvergenceWarning: Could not trim params automatically due to failed QC 
check. Trimming using trim_mode == 'size' will still work.
   warnings.warn(msg, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:71: 
ConvergenceWarning: QC check did not pass for 4 out of 6 parameters
Try increasing solver accuracy or number of iterations, decreasing 
alpha, or switch solvers
   warnings.warn(message, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:144: 
ConvergenceWarning: Could not trim params automatically due to failed QC 
check. Trimming using trim_mode == 'size' will still work.
   warnings.warn(msg, ConvergenceWarning)
 >>> res_reg.params
array([ 0.06225332, -0.64293349, -0.08217681,  0.00856738, -0.02679517,
         1.48236785])
 >>> 
alpha=np.ones(6);alpha[-2:]=0;res_reg=a.res1.model.fit_regularized(alpha=alpha*1e-3,disp=0,maxiter=500)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:71: 
ConvergenceWarning: QC check did not pass for 4 out of 4 parameters
Try increasing solver accuracy or number of iterations, decreasing 
alpha, or switch solvers
   warnings.warn(message, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:144: 
ConvergenceWarning: Could not trim params automatically due to failed QC 
check. Trimming using trim_mode == 'size' will still work.
   warnings.warn(msg, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:71: 
ConvergenceWarning: QC check did not pass for 4 out of 6 parameters
Try increasing solver accuracy or number of iterations, decreasing 
alpha, or switch solvers
   warnings.warn(message, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:144: 
ConvergenceWarning: Could not trim params automatically due to failed QC 
check. Trimming using trim_mode == 'size' will still work.
   warnings.warn(msg, ConvergenceWarning)
 >>> 
alpha=np.ones(6);alpha[-2:]=0;res_reg=a.res1.model.fit_regularized(alpha=alpha*1e-1,disp=0,maxiter=500)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:71: 
ConvergenceWarning: QC check did not pass for 2 out of 4 parameters
Try increasing solver accuracy or number of iterations, decreasing 
alpha, or switch solvers
   warnings.warn(message, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:144: 
ConvergenceWarning: Could not trim params automatically due to failed QC 
check. Trimming using trim_mode == 'size' will still work.
   warnings.warn(msg, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:71: 
ConvergenceWarning: QC check did not pass for 2 out of 6 parameters
Try increasing solver accuracy or number of iterations, decreasing 
alpha, or switch solvers
   warnings.warn(message, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:144: 
ConvergenceWarning: Could not trim params automatically due to failed QC 
check. Trimming using trim_mode == 'size' will still work.
   warnings.warn(msg, ConvergenceWarning)
 >>> res_reg.params
array([ 0.06224928, -0.64291724, -0.08216221,  0.00856535, -0.02679489,
         1.48237496])
 >>>

amd64:

 >>> 
alpha=np.ones(6);alpha[-2:]=0;res_reg=a.res1.model.fit_regularized(alpha=alpha*0.01,disp=0,maxiter=500)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:71: 
ConvergenceWarning: QC check did not pass for 4 out of 4 parameters
Try increasing solver accuracy or number of iterations, decreasing 
alpha, or switch solvers
   warnings.warn(message, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:144: 
ConvergenceWarning: Could not trim params automatically due to failed QC 
check. Trimming using trim_mode == 'size' will still work.
   warnings.warn(msg, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:71: 
ConvergenceWarning: QC check did not pass for 3 out of 6 parameters
Try increasing solver accuracy or number of iterations, decreasing 
alpha, or switch solvers
   warnings.warn(message, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:144: 
ConvergenceWarning: Could not trim params automatically due to failed QC 
check. Trimming using trim_mode == 'size' will still work.
   warnings.warn(msg, ConvergenceWarning)
 >>> alpha
array([1., 1., 1., 1., 0., 0.])
 >>> a.res1.params               array([ 0.06225337, -0.6429324 , 
-0.0821788 ,  0.00856726, -0.02679518,
         1.4823691 ])
 >>> res_reg.params
array([ 0.06225553, -0.64293539, -0.08217582,  0.00856681, -0.02679528,
         1.48236907])
 >>> a.res1.exog_names
Traceback (most recent call last):
   File "<stdin>", line 1, in <module>
   File "/usr/lib/python3/dist-packages/statsmodels/base/wrapper.py", 
line 36, in __getattribute__
     obj = getattr(results, attr)
AttributeError: 'ZeroInflatedPoissonResults' object has no attribute 
'exog_names'
 >>> a.res1.model.exog_names
['inflate_x1', 'inflate_const', 'x1', 'x2', 'x3', 'const']
 >>> 
alpha=np.ones(6);alpha[-2:]=0;res_reg=a.res1.model.fit_regularized(alpha=alpha*0.01,disp=0,maxiter=500)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:71: 
ConvergenceWarning: QC check did not pass for 4 out of 4 parameters
Try increasing solver accuracy or number of iterations, decreasing 
alpha, or switch solvers
   warnings.warn(message, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:144: 
ConvergenceWarning: Could not trim params automatically due to failed QC 
check. Trimming using trim_mode == 'size' will still work.
   warnings.warn(msg, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:71: 
ConvergenceWarning: QC check did not pass for 3 out of 6 parameters
Try increasing solver accuracy or number of iterations, decreasing 
alpha, or switch solvers
   warnings.warn(message, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:144: 
ConvergenceWarning: Could not trim params automatically due to failed QC 
check. Trimming using trim_mode == 'size' will still work.
   warnings.warn(msg, ConvergenceWarning)
 >>> 
alpha=np.ones(6);alpha[-2:]=0;res_reg=a.res1.model.fit_regularized(alpha=alpha*0.0001,disp=0,maxiter=500)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:71: 
ConvergenceWarning: QC check did not pass for 4 out of 4 parameters
Try increasing solver accuracy or number of iterations, decreasing 
alpha, or switch solvers
   warnings.warn(message, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:144: 
ConvergenceWarning: Could not trim params automatically due to failed QC 
check. Trimming using trim_mode == 'size' will still work.
   warnings.warn(msg, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:71: 
ConvergenceWarning: QC check did not pass for 4 out of 6 parameters
Try increasing solver accuracy or number of iterations, decreasing 
alpha, or switch solvers
   warnings.warn(message, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:144: 
ConvergenceWarning: Could not trim params automatically due to failed QC 
check. Trimming using trim_mode == 'size' will still work.
   warnings.warn(msg, ConvergenceWarning)
 >>> 
alpha=np.ones(6);alpha[-2:]=0;res_reg=a.res1.model.fit_regularized(alpha=alpha*0.000001,disp=0,maxiter=500)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:71: 
ConvergenceWarning: QC check did not pass for 4 out of 4 parameters
Try increasing solver accuracy or number of iterations, decreasing 
alpha, or switch solvers
   warnings.warn(message, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:144: 
ConvergenceWarning: Could not trim params automatically due to failed QC 
check. Trimming using trim_mode == 'size' will still work.
   warnings.warn(msg, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:71: 
ConvergenceWarning: QC check did not pass for 4 out of 6 parameters
Try increasing solver accuracy or number of iterations, decreasing 
alpha, or switch solvers
   warnings.warn(message, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:144: 
ConvergenceWarning: Could not trim params automatically due to failed QC 
check. Trimming using trim_mode == 'size' will still work.
   warnings.warn(msg, ConvergenceWarning)
 >>> 
alpha=np.ones(6);alpha[-2:]=0;res_reg=a.res1.model.fit_regularized(alpha=alpha*0.0,disp=0,maxiter=500)
 >>> res_reg.params
array([ 0.06225335, -0.64293223, -0.08217877,  0.00856726, -0.02679523,
         1.48236917])
 >>> a.res1.params
array([ 0.06225337, -0.6429324 , -0.0821788 ,  0.00856726, -0.02679518,
         1.4823691 ])
 >>> 
alpha=np.ones(6);alpha[-2:]=0;res_reg=a.res1.model.fit_regularized(alpha=alpha*1e-8,disp=0,maxiter=500)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:71: 
ConvergenceWarning: QC check did not pass for 4 out of 4 parameters
Try increasing solver accuracy or number of iterations, decreasing 
alpha, or switch solvers
   warnings.warn(message, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:144: 
ConvergenceWarning: Could not trim params automatically due to failed QC 
check. Trimming using trim_mode == 'size' will still work.
   warnings.warn(msg, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:71: 
ConvergenceWarning: QC check did not pass for 4 out of 6 parameters
Try increasing solver accuracy or number of iterations, decreasing 
alpha, or switch solvers
   warnings.warn(message, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:144: 
ConvergenceWarning: Could not trim params automatically due to failed QC 
check. Trimming using trim_mode == 'size' will still work.
   warnings.warn(msg, ConvergenceWarning)
 >>> 
alpha=np.ones(6);alpha[-2:]=0;res_reg=a.res1.model.fit_regularized(alpha=alpha*1e-14,disp=0,maxiter=500)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:71: 
ConvergenceWarning: QC check did not pass for 4 out of 4 parameters
Try increasing solver accuracy or number of iterations, decreasing 
alpha, or switch solvers
   warnings.warn(message, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:144: 
ConvergenceWarning: Could not trim params automatically due to failed QC 
check. Trimming using trim_mode == 'size' will still work.
   warnings.warn(msg, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:71: 
ConvergenceWarning: QC check did not pass for 4 out of 6 parameters
Try increasing solver accuracy or number of iterations, decreasing 
alpha, or switch solvers
   warnings.warn(message, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:144: 
ConvergenceWarning: Could not trim params automatically due to failed QC 
check. Trimming using trim_mode == 'size' will still work.
   warnings.warn(msg, ConvergenceWarning)
 >>> 
alpha=np.ones(6);alpha[-2:]=0;res_reg=a.res1.model.fit_regularized(alpha=alpha*1,disp=0,maxiter=500)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:71: 
ConvergenceWarning: QC check did not pass for 2 out of 4 parameters
Try increasing solver accuracy or number of iterations, decreasing 
alpha, or switch solvers
   warnings.warn(message, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:144: 
ConvergenceWarning: Could not trim params automatically due to failed QC 
check. Trimming using trim_mode == 'size' will still work.
   warnings.warn(msg, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:71: 
ConvergenceWarning: QC check did not pass for 1 out of 6 parameters
Try increasing solver accuracy or number of iterations, decreasing 
alpha, or switch solvers
   warnings.warn(message, ConvergenceWarning)
/usr/lib/python3/dist-packages/statsmodels/base/l1_solvers_common.py:144: 
ConvergenceWarning: Could not trim params automatically due to failed QC 
check. Trimming using trim_mode == 'size' will still work.
   warnings.warn(msg, ConvergenceWarning)
 >>> res_reg.params
array([ 0.06218367, -0.64270323, -0.08204927,  0.00855978, -0.02679348,
         1.48237349])
 >>> 
alpha=np.ones(6);alpha[-2:]=0;res_reg=a.res1.model.fit_regularized(alpha=alpha*1e3,disp=0,maxiter=500)
 >>> res_reg.params
array([ 0.        , -0.4368156 ,  0.        ,  0.00249224, -0.02523679,
         1.48930201])
 >>> alpha
array([1., 1., 1., 1., 0., 0.])
 >>> a.res1.params
array([ 0.06225337, -0.6429324 , -0.0821788 ,  0.00856726, -0.02679518,
         1.4823691 ])
 >>> 
alpha=np.ones(6);alpha[-2:]=0;res_reg=a.res1.model.fit_regularized(alpha=alpha*10,disp=0,maxiter=500)
 >>> res_reg.params
array([ 0.06155535, -0.64064802, -0.08088813,  0.00849131, -0.02677657,
         1.48241694])
 >>> 
alpha=np.ones(6);alpha[-2:]=0;res_reg=a.res1.model.fit_regularized(alpha=alpha*100,disp=0,maxiter=500)
 >>> res_reg.params
array([ 0.05531263, -0.62025736, -0.06936749,  0.00781262, -0.02661253,
         1.48282804])