MAINT: Remove appveyor and downscale travis

Fix docs to reflect changes
Down scale Travis to just docs

Author: Kevin Sheppard

.travis.yml.back renamed to .travis.yml

@@ -43,53 +43,15 @@ jobs:
   fast_finish: true
   include:
     # Documentation build (on Python 3.7 + cutting edge packages). Slowest build
-#    - python: 3.8
-#      env:
-#        - BUILD_INIT=tools/ci/travis_pip.sh
-#        - USE_MATPLOTLIB=false
-#        - USE_CVXOPT=false
-#        - LINT=true
-    - python: 3.7
+    - python: 3.8
       env:
         - PYTHON=${TRAVIS_PYTHON_VERSION}
         - DOCBUILD=true
-    # Python 3.7 + fixed pandas
-#    - python: 3.7
-#      env:
-#        - PYTHON=${TRAVIS_PYTHON_VERSION}
-#        - PANDAS=0.25
-#        - NUMPY=1.17
-#        - SCIPY=1.4
-#        - COVERAGE=true
-#        - PYTEST_OPTIONS=
-#    # Python 3.6 + legacy blas + older pandas
-#    - python: 3.6
-#      env:
-#        - PYTHON=${TRAVIS_PYTHON_VERSION}
-#        - NUMPY=1.16
-#        - PANDAS=0.24
-#        - SCIPY=1.3
-#        - BLAS="nomkl blas=*=openblas"
-#        - COVERAGE=true
-#        - PYTEST_OPTIONS=
-#    # Python 3.6 + oldest packages
-#    - python: 3.6
-#      env:
-#        - PYTHON=${TRAVIS_PYTHON_VERSION}
-#        - NUMPY=1.15
-#        - SCIPY=1.2
-#        - PANDAS=0.23
-#        - MATPLOTLIB=2
-#        - LINT=true
-#    # Latest pre-release packages
+    # Latest pre-release packages
     - python: 3.8
       env:
         - PIP_PRE=true
         - BUILD_INIT=tools/ci/travis_pip.sh
-#    - os: osx
-#      language: generic
-#      env:
-#        - PYTHON=3.7
 
   allow_failures:
     # pre-testing is a little fragile. Make it an FYI.

CONTRIBUTING.rst

@@ -50,7 +50,7 @@ So you want to submit a patch to `statsmodels` but are not too familiar with git
 
 1. `Fork <https://help.github.com/articles/fork-a-repo>`_ the `statsmodels repository <https://github.com/statsmodels/statsmodels>`_ on Github.
 2. `Create a new feature branch <https://git-scm.com/book/en/Git-Branching-Basic-Branching-and-Merging>`_. Each branch must be self-contained, with a single new feature or bugfix.
-3. Make sure the test suite passes. This includes testing on Python 3. The easiest way to do this is to either enable `Travis-CI <https://travis-ci.org/>`_ on your fork, or to make a pull request and check there.
+3. Make sure the test suite passes. This includes testing on Python 3. The easiest way to do this is to make a pull request and let the bot check for you. This can be slow, and if you are unsure about the fix or enhancement, it is best to run pytest locally.
 4. Document your changes by editing the appropriate file in ``docs/source/``. If it is a big, new feature add a note and an example to the latest ``docs/source/release/versionX.X.rst`` file. See older versions for examples. If it's a minor change, it will be included automatically in our release notes.
 5. Add an example. If it is a big, new feature please submit an example notebook by following `these instructions <https://www.statsmodels.org/devel/dev/examples.html>`_.
 6. `Submit a pull request <https://help.github.com/articles/using-pull-requests>`_

README.rst

@@ -1,5 +1,5 @@
-|PyPI Version| |Conda Version| |License| |Travis Build Status| |Azure CI Build Status|
-|Appveyor Build Status| |Coveralls Coverage| |PyPI downloads| |Conda downloads|
+|PyPI Version| |Conda Version| |License| |Azure CI Build Status|
+|Coveralls Coverage| |PyPI downloads| |Conda downloads|
 
 About statsmodels
 =================
@@ -174,12 +174,8 @@ Bug reports can be submitted to the issue tracker at
 
 https://github.com/statsmodels/statsmodels/issues
 
-.. |Travis Build Status| image:: https://travis-ci.org/statsmodels/statsmodels.svg?branch=master
-   :target: https://travis-ci.org/statsmodels/statsmodels
 .. |Azure CI Build Status| image:: https://dev.azure.com/statsmodels/statsmodels-testing/_apis/build/status/statsmodels.statsmodels?branch=master
    :target: https://dev.azure.com/statsmodels/statsmodels-testing/_build/latest?definitionId=1&branch=master
-.. |Appveyor Build Status| image:: https://ci.appveyor.com/api/projects/status/gx18sd2wc63mfcuc/branch/master?svg=true
-   :target: https://ci.appveyor.com/project/josef-pkt/statsmodels/branch/master
 .. |Coveralls Coverage| image:: https://coveralls.io/repos/github/statsmodels/statsmodels/badge.svg?branch=master
    :target: https://coveralls.io/github/statsmodels/statsmodels?branch=master
 .. |PyPI downloads| image:: https://img.shields.io/pypi/dm/statsmodels.svg?label=Pypi%20downloads

docs/source/about.rst

@@ -73,7 +73,7 @@ We would also like to thank our hosting providers, `github
 and `python.org <https://www.python.org/>`_ for making our downloads available
 on PyPi.
 
-We also thank our continuous integration providers,
+We also thank our current and previous continuous integration providers, `Azure <https://azure.microsoft.com/>`_,
 `Travis CI <https://travis-ci.org/>`_ and `AppVeyor <https://ci.appveyor.com>`_ for
 unit testing, and `Codecov <https://codecov.io>`_ and `Coveralls <https://coveralls.io>`_ for
 code coverage.

docs/source/dev/index.rst

@@ -68,8 +68,9 @@ github? Here are the steps you need to take.
 2. `Create a new feature branch <https://git-scm.com/book/en/v2/Git-Branching-Basic-Branching-and-Merging>`_.
    Each branch must be self-contained, with a single new feature or bugfix.
 3. Make sure the test suite passes. This includes testing on Python 3. The
-   easiest way to do this is to either enable `Travis-CI <https://travis-ci.org/>`_
-   on your fork, or to make a pull request and check there.
+   easiest way to do this is to make a pull request and let the bot check for
+   you. This can be slow, and if you are unsure about the fix or enhancement,
+   it is best to run pytest locally.
 4. `Submit a pull request <https://help.github.com/articles/about-pull-requests/>`_
 
 Pull requests are thoroughly reviewed before being accepted into the codebase.

docs/source/dev/testing.rst

@@ -6,8 +6,8 @@ tested against different dependency and Python versions and architectures.
 Check these logs periodically, make sure everything looks okay, and fix any
 failures:
 
-* `Travis CI <https://travis-ci.org/statsmodels/statsmodels/builds>`_
-* `Appveyor <https://ci.appveyor.com/project/josef-pkt/statsmodels/branch/master>`_
+* `Azure Pipelines <https://dev.azure.com/statsmodels/statsmodels-testing/_build?definitionId=1&_a=summary>`_
+
 
 The test coverage pages are here:
 

statsmodels/discrete/tests/test_constrained.py

@@ -135,7 +135,7 @@ def setup_class(cls):
         # example without offset
         formula = 'deaths ~ logpyears + smokes + C(agecat)'
         mod = Poisson.from_formula(formula, data=data)
-        # get start_params, example fails to converge on one py TravisCI
+        # get start_params, example fails to converge on one CI run
         k_vars = len(mod.exog_names)
         start_params = np.zeros(k_vars)
         start_params[0] = np.log(mod.endog.mean())
@@ -248,7 +248,7 @@ def setup_class(cls):
         formula = 'deaths ~ logpyears + smokes + C(agecat)'
         mod = Poisson.from_formula(formula, data=data)
 
-        # get start_params, example fails to converge on one py TravisCI
+        # get start_params, example fails to converge on one CI run
         k_vars = len(mod.exog_names)
         start_params = np.zeros(k_vars)
         start_params[0] = np.log(mod.endog.mean())

statsmodels/imputation/tests/test_bayes_mi.py

@@ -84,7 +84,7 @@ def model_args_fn(x):
         r = mi.fit()
         r.summary()  # smoke test
         # TODO: why does the test tolerance need to be so slack?
-        # There is unexpected variation across versions on travis.
+        # There is unexpected variation across versions
         assert_allclose(r.params, np.r_[
             -0.05347919, -0.02479701, 0.10075517], 0.25, 0)
 
@@ -163,7 +163,7 @@ def model_kwds_fn(x):
     r = mi.fit(results_cb=results_cb)
     r.summary()  # smoke test
     # TODO: why does the test tolerance need to be so slack?
-    # There is unexpected variation across versions on travis.
+    # There is unexpected variation across versions
     assert_allclose(r.params, np.r_[
             -0.05347919, -0.02479701, 0.10075517], 0.25, 0)
 

statsmodels/sandbox/regression/gmm.py

@@ -1721,7 +1721,7 @@ def __init__(self, endog, exog, instrument, **kwds):
 
     def fitstart(self):
         #todo: replace with or add call to distfn._fitstart
-        #      added but not used during testing, avoid Travis
+        #      added but not used during testing
         distfn = self.distfn
         if hasattr(distfn, '_fitstart'):
             start = distfn._fitstart(self.endog)

tools/ci/azure/azure_template_posix.yml

@@ -145,3 +145,9 @@ jobs:
   - bash: bash <(curl -s https://codecov.io/bash)
     displayName: 'CodeCov upload'
     condition: and(eq(variables['coverage'], 'true'), ne(variables['test.install'], 'true'))
+    continueOnError: true
+
+  - script: coveralls
+    displayName: 'coveralls upload'
+    condition: and(eq(variables['coverage'], 'true'), ne(variables['test.install'], 'true'))
+    continueOnError: true

tools/ci/azure/install-posix.sh

@@ -15,7 +15,7 @@ else
 fi
 
 python -m pip install --upgrade pip setuptools wheel
-python -m pip install cython pytest pytest-xdist coverage pytest-cov ipython jupyter notebook nbconvert "property_cached>=1.6.3" black==20.8b1 isort flake8 nbconvert==5.6.1
+python -m pip install cython pytest pytest-xdist coverage pytest-cov ipython jupyter notebook nbconvert "property_cached>=1.6.3" black==20.8b1 isort flake8 nbconvert==5.6.1 coveralls
 
 if [[ -n ${NUMPY} ]]; then CMD="$CMD==${NUMPY}"; fi;
 CMD="$CMD scipy"