Merge a882f6e into b12ff9b

.coveragerc

@@ -5,4 +5,4 @@ include = */pyramid/*
 omit =
     */setup.py
     */__check_build/*
-    */_build_utils/*
+    */_build_utils/*

.gitignore

@@ -9,6 +9,9 @@
 # Pypi pws
 .pypipws
 
+# Coverage
+coverage
+
 # OAuth access for automating releases
 ACCESS_TOKEN
 

.travis.yml

@@ -11,7 +11,8 @@ cache:
   - $HOME/.cache/pip
   - $HOME/.ccache
 
-before_install: source build_tools/travis/before_install.sh
+before_install:
+  - source build_tools/travis/before_install.sh
 env:
   global:
     - PYMODULE=pyramid
@@ -93,8 +94,13 @@ matrix:
 install: source build_tools/travis/install.sh
 before_script: bash build_tools/travis/before_script.sh
 script: bash build_tools/travis/test_script.sh
+
+# we set +e to re-instate the error-tolerant state. The problem is Travis is broken on Mac OS
+# builds right now due to Issue 6307 (https://github.com/travis-ci/travis-ci/issues/6307)
+# and fails at the end of a successful build. This will allow the build to recover even if
+# a non-zero status code is encountered. (had `- set +e`)
 after_success:
-  - source build_tools/travis/after_success.sh
+  - source build_tools/travis/after_success.sh  # || echo "shell_session_update failed"
   # Build the wheels every time so we can debug
   - bash build_tools/travis/build_wheels.sh
   - ls dist/

AUTHORS.md

@@ -5,6 +5,7 @@ The following people have been core contributors to `pyramid`'s development:
   * [Taylor Smith](https://github.com/tgsmith61591)
   * [Gary Foreman](https://github.com/garyForeman)
   * [Charles Drotar](https://github.com/charlesdrotar)
+  * [Steven Hoelscher](https://github.com/shoelsch)
 
 __Please do not email the authors directly with questions or issues.__ Rather, use
 the [issues](https://github.com/tgsmith61591/pyramid/issues) page. Furthermore, issues

ISSUE_TEMPLATE.md

@@ -1,7 +1,7 @@
 <!-- Instructions For Filing a Bug: https://github.com/scikit-learn/scikit-learn/blob/master/CONTRIBUTING.md#filing-bugs -->
 
 #### Description
-<!-- Example: Joblib Error thrown when calling fit on LatentDirichletAllocation with evaluate_every > 0-->
+<!-- Example: ValueError raised when fitting very high-frequency data -->
 
 #### Steps/Code to Reproduce
 <!--

appveyor.yml

@@ -34,7 +34,12 @@ build_script:
   - cmd: C:\Miniconda.exe /S /D=C:\Py
   - SET PATH=C:\Py;C:\Py\Scripts;C:\Py\Library\bin;%PATH%
   - conda config --set always_yes yes
-  - conda update conda --quiet
+  # - conda update conda
+
+  # Debugging because Appveyor gives us no stack traces when this stuff breaks.
+  - ps: echo "Conda version:"
+  - conda --version
+  - ps: echo "Python version=$env:PYTHON_VERSION"
 
   # Things we'll need for build/test
   - ps: conda create -n testenv --yes python=$env:PYTHON_VERSION numpy scipy cython scikit-learn statsmodels pip nose patsy

benchmarks/benchmark_funcs.py

@@ -0,0 +1,57 @@
+# -*- coding: utf-8 -*-
+#
+# Benchmark various approaches to functions to speed things up.
+# ... hopefully.
+
+from __future__ import absolute_import
+
+import numpy as np
+
+import time
+
+
+def _do_time(func, n_iter=10, *args, **kwargs):
+    times = []
+    for _ in range(n_iter):
+        start = time.time()
+        func(*args, **kwargs)
+        times.append(time.time() - start)
+
+    times = np.asarray(times)
+    print("Completed %i iterations (avg=%.6f, min=%.6f, max=%.6f)"
+          % (n_iter, times.mean(), times.min(), times.max()))
+
+
+def benchmark_is_constant():
+    """This benchmarks the "is_constant" function from ``pyramid.arima.utils``.
+    This was added in 0.6.2.
+    """
+    # WINNER!
+    def is_const1(x):
+        """This is the version in Pyramid 0.6.2.
+
+        Parameters
+        ----------
+        x : np.ndarray
+            This is the array.
+        """
+        return (x == x[0]).all()
+
+    def is_const2(x):
+        """This should ostensibly only take O(N) rather than O(2N) like
+        its predecessor. But we'll see...
+
+        Parameters
+        ----------
+        x : np.ndarray
+            This is the array.
+        """
+        return np.unique(x).shape[0] == 1
+
+    x = np.random.choice(np.arange(10), 1000000, replace=True)
+    _do_time(is_const1, 25, x)
+    _do_time(is_const2, 25, x)
+
+
+if __name__ == '__main__':
+    benchmark_is_constant()

build_tools/travis/after_success.sh

@@ -4,10 +4,13 @@
 
 # License: 3-clause BSD
 
-set -e
+# 02/10/2018 remove due to Travis build issue 6307
+# set -e
+set +e  # because TRAVIS SUCKS
 
 # push coverage if necessary
 if [[ "$COVERAGE" == "true" ]]; then
+
     # Need to run coveralls from a git checkout, so we copy .coverage
     # from TEST_DIR where nosetests has been run
     cp $TEST_DIR/.coverage $TRAVIS_BUILD_DIR

build_tools/travis/before_install.sh

@@ -6,4 +6,15 @@ set -e
 if [[ "$TRAVIS_OS_NAME" == "linux" ]]; then
   echo "Updating apt-get for Linux build"
   sudo apt-get -qq update
+
+# Workaround for https://github.com/travis-ci/travis-ci/issues/6307, which
+# caused the following error on MacOS X workers:
+#
+# Warning, RVM 1.26.0 introduces signed releases and automated check of
+# signatures when GPG software found.
+# /Users/travis/build.sh: line 109: shell_session_update: command not found
+elif [[ "$TRAVIS_OS_NAME" == "osx" ]]; then
+  echo "Updating Ruby for Mac OS build"
+  # command curl -sSL https://rvm.io/mpapis.asc | gpg --import -;
+  # rvm get stable
 fi

doc/_static/css/gitcontrib.css

@@ -32,6 +32,11 @@
   margin-top: 10px;
 }
 
+img.avatar {
+  width: 60px;
+  height: 60px;
+}
+
 .contrib-author-wrapper {
   position: absolute;
   display: block;

doc/_static/js/contrib.js

@@ -2,7 +2,7 @@ function commaFmt(x) {
     return x.toString().replace(/\B(?=(\d{3})+(?!\d))/g, ",");
 }
 
-$(document).ready(function() {
+function fetchContributors() {
     $.getJSON("https://api.github.com/repos/tgsmith61591/pyramid/stats/contributors", function(arr) {
         // sort the array based on total count
         arr.sort(function(a, b) {
@@ -54,7 +54,8 @@ $(document).ready(function() {
                          '</div>' +
                         '</li>')
 
+            // can only do this once the doc is ready
             $('#contrib').append(li);
         });
     });
-});
+}

doc/contributors.rst

@@ -11,6 +11,12 @@ Thanks to the following users for their contributions to Pyramid!
 
     <!-- This does all the heavy lifting -->
     <script src="_static/js/contrib.js"></script>
+    <script type="text/javascript">
+        // actually call the contrib code
+        $(document).ready(function() {
+            fetchContributors();
+        });
+    </script>
 
     <!-- This is taken from the Github contrib page -->
     <ol id="contrib" class="contrib-data capped-cards clearfix"></ol>

doc/quickstart.rst

@@ -20,26 +20,26 @@ Here is a simple example of Pyramid use:
     ...                           error_action='ignore',  # don't want to know if an order does not work
     ...                           suppress_warnings=True,  # don't want convergence warnings
     ...                           stepwise=True)  # set to stepwise
-    Fit ARIMA: order=(1, 1, 1) seasonal_order=(0, 1, 1, 12); AIC=3066.811, BIC=3082.663, Fit time=1.731 seconds
+    Fit ARIMA: order=(1, 1, 1) seasonal_order=(0, 1, 1, 12); AIC=3066.811, BIC=3082.663, Fit time=0.517 seconds
     Fit ARIMA: order=(0, 1, 0) seasonal_order=(0, 1, 0, 12); AIC=nan, BIC=nan, Fit time=nan seconds
-    Fit ARIMA: order=(1, 1, 0) seasonal_order=(1, 1, 0, 12); AIC=3099.735, BIC=3112.417, Fit time=0.434 seconds
-    Fit ARIMA: order=(0, 1, 1) seasonal_order=(0, 1, 1, 12); AIC=3066.983, BIC=3079.665, Fit time=0.473 seconds
-    Fit ARIMA: order=(1, 1, 1) seasonal_order=(1, 1, 1, 12); AIC=3067.666, BIC=3086.688, Fit time=2.184 seconds
-    Fit ARIMA: order=(1, 1, 1) seasonal_order=(0, 1, 0, 12); AIC=3088.109, BIC=3100.791, Fit time=0.344 seconds
-    Fit ARIMA: order=(1, 1, 1) seasonal_order=(0, 1, 2, 12); AIC=3067.669, BIC=3086.692, Fit time=4.371 seconds
-    Fit ARIMA: order=(1, 1, 1) seasonal_order=(1, 1, 2, 12); AIC=3068.757, BIC=3090.951, Fit time=4.160 seconds
-    Fit ARIMA: order=(2, 1, 1) seasonal_order=(0, 1, 1, 12); AIC=3067.485, BIC=3086.508, Fit time=1.204 seconds
-    Fit ARIMA: order=(1, 1, 0) seasonal_order=(0, 1, 1, 12); AIC=3094.578, BIC=3107.260, Fit time=0.468 seconds
-    Fit ARIMA: order=(1, 1, 2) seasonal_order=(0, 1, 1, 12); AIC=3066.771, BIC=3085.794, Fit time=1.222 seconds
-    Fit ARIMA: order=(2, 1, 3) seasonal_order=(0, 1, 1, 12); AIC=3070.642, BIC=3096.006, Fit time=3.495 seconds
-    Fit ARIMA: order=(1, 1, 2) seasonal_order=(1, 1, 1, 12); AIC=3068.086, BIC=3090.280, Fit time=1.180 seconds
-    Fit ARIMA: order=(1, 1, 2) seasonal_order=(0, 1, 0, 12); AIC=3090.977, BIC=3106.830, Fit time=0.544 seconds
-    Fit ARIMA: order=(1, 1, 2) seasonal_order=(0, 1, 2, 12); AIC=3067.766, BIC=3089.959, Fit time=3.303 seconds
-    Fit ARIMA: order=(1, 1, 2) seasonal_order=(1, 1, 2, 12); AIC=3069.717, BIC=3095.081, Fit time=5.675 seconds
+    Fit ARIMA: order=(1, 1, 0) seasonal_order=(1, 1, 0, 12); AIC=3099.735, BIC=3112.417, Fit time=0.162 seconds
+    Fit ARIMA: order=(0, 1, 1) seasonal_order=(0, 1, 1, 12); AIC=3066.983, BIC=3079.665, Fit time=0.164 seconds
+    Fit ARIMA: order=(1, 1, 1) seasonal_order=(1, 1, 1, 12); AIC=3067.666, BIC=3086.688, Fit time=0.645 seconds
+    Fit ARIMA: order=(1, 1, 1) seasonal_order=(0, 1, 0, 12); AIC=3088.109, BIC=3100.791, Fit time=0.136 seconds
+    Fit ARIMA: order=(1, 1, 1) seasonal_order=(0, 1, 2, 12); AIC=3067.669, BIC=3086.692, Fit time=1.512 seconds
+    Fit ARIMA: order=(1, 1, 1) seasonal_order=(1, 1, 2, 12); AIC=3068.757, BIC=3090.951, Fit time=1.651 seconds
+    Fit ARIMA: order=(2, 1, 1) seasonal_order=(0, 1, 1, 12); AIC=3067.485, BIC=3086.508, Fit time=0.445 seconds
+    Fit ARIMA: order=(1, 1, 0) seasonal_order=(0, 1, 1, 12); AIC=3094.578, BIC=3107.260, Fit time=0.174 seconds
+    Fit ARIMA: order=(1, 1, 2) seasonal_order=(0, 1, 1, 12); AIC=3066.771, BIC=3085.794, Fit time=0.425 seconds
+    Fit ARIMA: order=(2, 1, 3) seasonal_order=(0, 1, 1, 12); AIC=3070.642, BIC=3096.006, Fit time=0.966 seconds
+    Fit ARIMA: order=(1, 1, 2) seasonal_order=(1, 1, 1, 12); AIC=3068.086, BIC=3090.280, Fit time=0.411 seconds
+    Fit ARIMA: order=(1, 1, 2) seasonal_order=(0, 1, 0, 12); AIC=3090.977, BIC=3106.830, Fit time=0.249 seconds
+    Fit ARIMA: order=(1, 1, 2) seasonal_order=(0, 1, 2, 12); AIC=3067.766, BIC=3089.959, Fit time=1.170 seconds
+    Fit ARIMA: order=(1, 1, 2) seasonal_order=(1, 1, 2, 12); AIC=3069.717, BIC=3095.081, Fit time=2.000 seconds
     Fit ARIMA: order=(0, 1, 2) seasonal_order=(0, 1, 1, 12); AIC=nan, BIC=nan, Fit time=nan seconds
-    Fit ARIMA: order=(2, 1, 2) seasonal_order=(0, 1, 1, 12); AIC=3068.701, BIC=3090.895, Fit time=1.509 seconds
-    Fit ARIMA: order=(1, 1, 3) seasonal_order=(0, 1, 1, 12); AIC=3068.842, BIC=3091.036, Fit time=2.329 seconds
-    Total fit time: 34.636 seconds
+    Fit ARIMA: order=(2, 1, 2) seasonal_order=(0, 1, 1, 12); AIC=3068.701, BIC=3090.895, Fit time=0.523 seconds
+    Fit ARIMA: order=(1, 1, 3) seasonal_order=(0, 1, 1, 12); AIC=3068.842, BIC=3091.036, Fit time=0.590 seconds
+    Total fit time: 11.745 seconds
 
 It's easy to examine your model fit results. Simply use the ``summary`` method:
 

doc/setup.rst

@@ -17,3 +17,6 @@ python:
 .. code-block:: python
 
     from pyramid.arima import auto_arima
+
+If you encounter an ``ImportError``, try updating numpy and re-installing. Outdated
+numpy versions have been observed to break the Pyramid build.

pyramid/__init__.py

@@ -4,7 +4,7 @@
 #
 # The pyramid module
 
-__version__ = "0.6.2"
+__version__ = "0.6.5"
 
 try:
     # this var is injected in the setup build to enable

pyramid/arima/__init__.py

@@ -6,5 +6,6 @@
 from .arima import *
 from .auto import *
 from .utils import *
+from .warnings import *
 
 __all__ = [s for s in dir() if not s.startswith("_")]

pyramid/arima/approx.py

@@ -33,6 +33,10 @@
     'mean': np.average
 }
 
+# identity function defined once to avoid multiple lambda calls
+# littered throughout
+_identity = (lambda t: t)
+
 
 def _regularize(x, y, ties):
     """Regularize the values, make them ordered and remove duplicates.
@@ -82,8 +86,14 @@ def tie_apply(f, u_val):
                 return f(vals)
 
             # replace the duplicates in the y array with the "tie" func
-            func = VALID_TIES.get(ties, lambda t: t)
+            func = VALID_TIES.get(ties, _identity)
+
+            # maybe expensive to vectorize on the fly? Not sure; would need
+            # to do some benchmarking. However, we need to in order to keep y
+            # and x in scope...
             y = np.vectorize(tie_apply)(func, ux)
+
+            # does ux need ordering? hmm..
             x = ux
 
     return x, y

pyramid/arima/arima.py

@@ -215,6 +215,9 @@ def fit(self, y, exogenous=None, **fit_args):
             operation. This should not include a constant or trend. Note that
             if an ``ARIMA`` is fit on exogenous features, it must be provided
             exogenous features for making predictions.
+
+        **fit_args : dict or kwargs
+            Any keyword arguments to pass to the statsmodels ARIMA fit.
         """
         y = c1d(check_array(y, ensure_2d=False, force_all_finite=False,
                             copy=True, dtype=DTYPE))  # type: np.ndarray
@@ -360,9 +363,8 @@ def predict_in_sample(self, exogenous=None, start=None,
 
     def predict(self, n_periods=10, exogenous=None):
         """Generate predictions (forecasts) ``n_periods`` in the future.
-        Note that unless ``include_std_err`` or ``include_conf_int`` are True,
-        only the forecast array will be returned (otherwise, a tuple with the
-        corresponding elements will be returned).
+        Note that if ``exogenous`` variables were used in the model fit, they
+        will be expected for the predict procedure and will fail otherwise.
 
         Parameters
         ----------
@@ -423,7 +425,7 @@ def fit_predict(self, y, exogenous=None, n_periods=10, **fit_args):
         n_periods : int, optional (default=10)
             The number of periods in the future to forecast.
 
-        fit_args : dict, optional (default=None)
+        fit_args : dict or kwargs, optional (default=None)
             Any keyword args to pass to the fit method.
         """
         self.fit(y, exogenous, **fit_args)