Error in Predict Class Probability for SVC and Linear SVC

[Agustin Lobo]

In the case of SVC, the Predict Class Probability stops with error:

AttributeError: predict_proba is not available when probability=False

This is also the case if using the Classification Workflow, with a slightly different message:

Error: predict_proba is not available when probability=False

‌

[Fabian Thiel]

EDIT

@Agustin Lobo Try setting the parameter probability=False to probability=True in the code section of Fit SVC, then it should work!

svc = SVC(probability=True)

‌

‌

@Andreas Janz I can reproduce this with the testdata:

Also have a look at #326

‌

QGIS version: 3.10.0-A Coruña
QGIS code revision: 6c816b4204
Qt version: 5.11.2
GDAL version: 2.4.1
GEOS version: 3.8.0-CAPI-1.13.1 
PROJ version: Rel. 5.2.0, September 15th, 2018
Processing algorithm…
Algorithm 'Predict Class Probability' starting…
Input parameters:
{ 'classifier' : 'C:/Users/ThinkPad/AppData/Roaming/QGIS/QGIS3/profiles/default/processing/outputs/outEstimator.pkl', 'mask' : None, 'outFraction' : 'TEMPORARY_OUTPUT', 'raster' : 'C:/Users/ThinkPad/AppData/Roaming/QGIS/QGIS3/profiles/default/python/plugins/enmapboxplugin/enmapboxtestdata/enmap_berlin.bsq' }

start EstimatorPredictProbability, RasterSize(x=220, y=400)
Traceback (most recent call last):
File "C:\Users\ThinkPad\AppData\Roaming\QGIS\QGIS3\profiles\default\python\plugins\enmapboxplugin\site-packages\enmapboxgeoalgorithms\provider.py", line 151, in processAlgorithm
result = self.processAlgorithm_()
File "C:\Users\ThinkPad\AppData\Roaming\QGIS\QGIS3\profiles\default\python\plugins\enmapboxplugin\site-packages\enmapboxgeoalgorithms\algorithms.py", line 394, in processAlgorithm_
estimator.predictProbability(filename=filename, raster=raster, mask=mask, progressBar=self._progressBar)
File "C:\Users\ThinkPad\AppData\Roaming\QGIS\QGIS3\profiles\default\python\plugins\enmapboxplugin\site-packages\hubflow\core.py", line 5238, in _predictProbability
applier.apply(operatorType=_EstimatorPredictProbability, raster=raster, estimator=self, mask=mask, mask2=mask2)
File "C:\Users\ThinkPad\AppData\Roaming\QGIS\QGIS3\profiles\default\python\plugins\enmapboxplugin\site-packages\hubflow\core.py", line 48, in apply
overwrite=self.kwargs.get('overwrite', True), *ufuncArgs, **ufuncKwargs)
File "C:\Users\ThinkPad\AppData\Roaming\QGIS\QGIS3\profiles\default\python\plugins\enmapboxplugin\site-packages\hubdc\applier.py", line 976, in apply
results = self._runProcessSubgrids()
File "C:\Users\ThinkPad\AppData\Roaming\QGIS\QGIS3\profiles\default\python\plugins\enmapboxplugin\site-packages\hubdc\applier.py", line 1040, in _runProcessSubgrids
blockResults.append(_Worker.processSubgrid(**kwargs))
File "C:\Users\ThinkPad\AppData\Roaming\QGIS\QGIS3\profiles\default\python\plugins\enmapboxplugin\site-packages\hubdc\applier.py", line 1123, in processSubgrid
nxblock=nx)
File "C:\Users\ThinkPad\AppData\Roaming\QGIS\QGIS3\profiles\default\python\plugins\enmapboxplugin\site-packages\hubdc\applier.py", line 1269, in _apply
blockResult = self.ufunc(*self._ufuncArgs, **self._ufuncKwargs)
File "C:\Users\ThinkPad\AppData\Roaming\QGIS\QGIS3\profiles\default\python\plugins\enmapboxplugin\site-packages\hubflow\core.py", line 5336, in ufunc
y = estimator.sklEstimator().predict_proba(X=X)
File "C:\OSGEO4~1\apps\Python37\lib\site-packages\sklearn\utils\metaestimators.py", line 110, in __get__
getattr(delegate, self.attribute_name)
File "C:\OSGEO4~1\apps\Python37\lib\site-packages\sklearn\utils\metaestimators.py", line 110, in __get__
getattr(delegate, self.attribute_name)
File "C:\OSGEO4~1\apps\Python37\lib\site-packages\sklearn\svm\base.py", line 616, in predict_proba
self._check_proba()
File "C:\OSGEO4~1\apps\Python37\lib\site-packages\sklearn\svm\base.py", line 583, in _check_proba
raise AttributeError("predict_proba is not available when "
AttributeError: predict_proba is not available when probability=False

Traceback (most recent call last):
File "C:\Users\ThinkPad\AppData\Roaming\QGIS\QGIS3\profiles\default\python\plugins\enmapboxplugin\site-packages\enmapboxgeoalgorithms\provider.py", line 151, in processAlgorithm
result = self.processAlgorithm_()
File "C:\Users\ThinkPad\AppData\Roaming\QGIS\QGIS3\profiles\default\python\plugins\enmapboxplugin\site-packages\enmapboxgeoalgorithms\algorithms.py", line 394, in processAlgorithm_
estimator.predictProbability(filename=filename, raster=raster, mask=mask, progressBar=self._progressBar)
File "C:\Users\ThinkPad\AppData\Roaming\QGIS\QGIS3\profiles\default\python\plugins\enmapboxplugin\site-packages\hubflow\core.py", line 5238, in _predictProbability
applier.apply(operatorType=_EstimatorPredictProbability, raster=raster, estimator=self, mask=mask, mask2=mask2)
File "C:\Users\ThinkPad\AppData\Roaming\QGIS\QGIS3\profiles\default\python\plugins\enmapboxplugin\site-packages\hubflow\core.py", line 48, in apply
overwrite=self.kwargs.get('overwrite', True), *ufuncArgs, **ufuncKwargs)
File "C:\Users\ThinkPad\AppData\Roaming\QGIS\QGIS3\profiles\default\python\plugins\enmapboxplugin\site-packages\hubdc\applier.py", line 976, in apply
results = self._runProcessSubgrids()
File "C:\Users\ThinkPad\AppData\Roaming\QGIS\QGIS3\profiles\default\python\plugins\enmapboxplugin\site-packages\hubdc\applier.py", line 1040, in _runProcessSubgrids
blockResults.append(_Worker.processSubgrid(**kwargs))
File "C:\Users\ThinkPad\AppData\Roaming\QGIS\QGIS3\profiles\default\python\plugins\enmapboxplugin\site-packages\hubdc\applier.py", line 1123, in processSubgrid
nxblock=nx)
File "C:\Users\ThinkPad\AppData\Roaming\QGIS\QGIS3\profiles\default\python\plugins\enmapboxplugin\site-packages\hubdc\applier.py", line 1269, in _apply
blockResult = self.ufunc(*self._ufuncArgs, **self._ufuncKwargs)
File "C:\Users\ThinkPad\AppData\Roaming\QGIS\QGIS3\profiles\default\python\plugins\enmapboxplugin\site-packages\hubflow\core.py", line 5336, in ufunc
y = estimator.sklEstimator().predict_proba(X=X)
File "C:\OSGEO4~1\apps\Python37\lib\site-packages\sklearn\utils\metaestimators.py", line 110, in __get__
getattr(delegate, self.attribute_name)
File "C:\OSGEO4~1\apps\Python37\lib\site-packages\sklearn\utils\metaestimators.py", line 110, in __get__
getattr(delegate, self.attribute_name)
File "C:\OSGEO4~1\apps\Python37\lib\site-packages\sklearn\svm\base.py", line 616, in predict_proba
self._check_proba()
File "C:\OSGEO4~1\apps\Python37\lib\site-packages\sklearn\svm\base.py", line 583, in _check_proba
raise AttributeError("predict_proba is not available when "
AttributeError: predict_proba is not available when probability=False

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "C:\Users\ThinkPad\AppData\Roaming\QGIS\QGIS3\profiles\default\python\plugins\enmapboxplugin\site-packages\enmapboxgeoalgorithms\provider.py", line 165, in processAlgorithm
raise Exception('unexpected error')
Exception: unexpected error

Execution failed after 0.96 seconds

Loading resulting layers
Algorithm 'Predict Class Probability' finished

‌

[Agustin Lobo]

I confirm it works by just setting

svc = SVC(probability=True)