Compatibility with Python 3.7.0b5 #11224
Description
When running the tests suite on Python 3.7.0b5, with
Dockerfile
FROM python:3.7-rc
RUN apt-get update && apt-get install -y build-essential git make
RUN apt-get install -y libopenblas-dev
RUN python --version
RUN pip3 --version
RUN pip3 install Cython nose pytest numpy
RUN pip3 install scipy
RUN git clone --depth 1 https://github.com/scikit-learn/scikit-learn.git && \
cd scikit-learn && git checkout master
RUN cd scikit-learn && pip3 install -e .
RUN cd scikit-learn && pytest sklearn
most tests appears to pass, but there are two minor doctest failures,
________________ [doctest] sklearn.exceptions.FitFailedWarning _________________
106 Examples
107 --------
108 >>> from sklearn.model_selection import GridSearchCV
109 >>> from sklearn.svm import LinearSVC
110 >>> from sklearn.exceptions import FitFailedWarning
111 >>> import warnings
112 >>> warnings.simplefilter('always', FitFailedWarning)
113 >>> gs = GridSearchCV(LinearSVC(), {'C': [-1, -2]}, error_score=0, cv=2)
114 >>> X, y = [[1, 2], [3, 4], [5, 6], [7, 8]], [0, 0, 1, 1]
115 >>> with warnings.catch_warnings(record=True) as w:
Expected:
FitFailedWarning('Estimator fit failed. The score on this train-test
partition for these parameters will be set to 0.000000.
Details: \nValueError: Penalty term must be positive; got (C=-2)\n',)
Got:
FitFailedWarning('Estimator fit failed. The score on this train-test partition for these parameters will be set to 0.000000. Details: \nValueError: Penalty term must be positive; got (C=-2)\n')
/scikit-learn/sklearn/exceptions.py:115: DocTestFailure
_________________ [doctest] sklearn.exceptions.NotFittedError __________________
019 Exception class to raise if estimator is used before fitting.
020
021 This class inherits from both ValueError and AttributeError to help with
022 exception handling and backward compatibility.
023
024 Examples
025 --------
026 >>> from sklearn.svm import LinearSVC
027 >>> from sklearn.exceptions import NotFittedError
028 >>> try:
Expected:
NotFittedError('This LinearSVC instance is not fitted yet',)
Got:
NotFittedError('This LinearSVC instance is not fitted yet')
Also there is a DeprecationWarning when running import sklearn reported in #11121