Merge pull request #13 from phobson/diy-plotting-positions

DIY plotting positions

.travis.yml

@@ -8,6 +8,9 @@ matrix:
     - python: 3.4
       env:
         - COVERAGE=false
+    - python: 3.5
+      env:
+        - COVERAGE=false
     - python: 3.5
       env:
         - COVERAGE=true
@@ -29,7 +32,8 @@ install:
 
   - conda create --yes -n test python=$TRAVIS_PYTHON_VERSION
   - source activate test
-  - conda install --yes numpy nose scipy matplotlib coverage docopt requests pyyaml
+  - conda install --yes numpy nose matplotlib coverage docopt requests pyyaml
+  - if [ ${COVERAGE} = true ]; then conda install scipy --yes; fi
   - pip install coveralls
   - pip install .
 

README.md

@@ -32,3 +32,15 @@ seaborn.despine(fig=fig)
 ```
 
 ![Alt text](docs/img/example.png "Example axes")
+
+## Testing
+
+Testing is generally done via the ``nose`` and ``numpy.testing`` modules.
+The best way to run the tests is in an interactive python session:
+
+```python
+import matplotlib
+matplotib.use('agg')
+import probscale
+probscale.test()
+```

conda.recipe/dev/meta.yaml

@@ -0,0 +1,44 @@
+package:
+  name: mpl-probscale
+  version: 0.1.1
+
+source:
+  path: ../../
+#  patches:
+   # List any patch files here
+   # - fix.patch
+
+build:
+  number: 1
+
+requirements:
+  build:
+    - python
+    - numpy
+    - matplotlib
+
+  run:
+    - python
+    - numpy
+    - matplotlib
+    - nose
+
+test:
+  imports:
+    - probscale
+
+  commands:
+    - python -c "import matplotlib; matplotlib.use('agg'); import probscale; probscale.test()"
+
+  requires:
+    - nose
+    - scipy
+
+about:
+  home: http://phobson.github.io/mpl-probscale/
+  license:  BSD License
+  summary: 'Probability scales for matplotlib.'
+
+# See
+# http://docs.continuum.io/conda/build.html for
+# more information about meta.yaml/configure

conda.recipe/release/bld.bat

@@ -0,0 +1,8 @@
+"%PYTHON%" setup.py install
+if errorlevel 1 exit 1
+
+:: Add more build steps here, if they are necessary.
+
+:: See
+:: http://docs.continuum.io/conda/build.html
+:: for a list of environment variables that are set during the build process.

conda.recipe/release/build.sh

@@ -0,0 +1,9 @@
+#!/bin/bash
+
+$PYTHON setup.py install
+
+# Add more build steps here, if they are necessary.
+
+# See
+# http://docs.continuum.io/conda/build.html
+# for a list of environment variables that are set during the build process.

conda.recipe/meta.yaml → conda.recipe/release/meta.yaml

@@ -1,10 +1,10 @@
 package:
   name: mpl-probscale
-  version: v0.1.0
+  version: 0.1.1
 
 source:
   git_url: https://github.com/phobson/mpl-probscale.git
-  git_tag: v0.1.0
+  git_tag: v0.1.1
 #  patches:
    # List any patch files here
    # - fix.patch
@@ -17,23 +17,19 @@ requirements:
     - python
     - numpy
     - matplotlib
-    - scipy
 
   run:
     - python
     - numpy
     - matplotlib
-    - scipy
+    - nose
 
 test:
   imports:
     - probscale
 
   commands:
-    - nosetests
-
-  requires:
-    - nose
+    - python -c "import matplotlib; matplotlib.use('agg'); import probscale; probscale.test()"
 
 about:
   home: http://phobson.github.io/mpl-probscale/

docs/conf.py

@@ -76,9 +76,9 @@
 # built documents.
 #
 # The short X.Y version.
-version = '0.1.0'
+version = '0.1.1'
 # The full version, including alpha/beta/rc tags.
-release = '0.1.0'
+release = '0.1.1'
 
 # The language for content autogenerated by Sphinx. Refer to documentation
 # for a list of supported languages.

docs/index.rst

@@ -54,22 +54,31 @@ Tutorial
 
    tutorial/getting_started.rst
 
+Testing
+-------
+
+It's easiest to run the tests from an interactive python session:
+
+.. code-block:: python
+
+    import matplotlib
+    matplotlib.use('agg')
+    import probscale
+    probscale.test()
+
+
 API References
 --------------
 
 .. toctree::
    :maxdepth: 2
 
-   probscale.rst
    viz.rst
-
-
-
+   probscale.rst
 
 Indices and tables
 ==================
 
 * :ref:`genindex`
 * :ref:`modindex`
 * :ref:`search`
-

docs/tutorial/getting_started.ipynb

@@ -4,7 +4,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# Getting started with `mpl-probscale`.\n",
+    "# Getting started with `mpl-probscale`\n",
     "\n",
     "## Installation\n",
     "\n",
@@ -21,7 +21,7 @@
     "pip install .\n",
     "```\n",
     "\n",
-    "I recommend `pip install .` over `python setup.py install` for [reasons I don't fully understand](https://github.com/numpy/numpy/issues/6551)."
+    "I recommend `pip install .` over `python setup.py install` for [reasons I don't fully understand](https://github.com/numpy/numpy/issues/6551#issuecomment-150392865cd)."
    ]
   },
   {
@@ -118,7 +118,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "`mpl-probscale` let's you use probability scales. All you need to do is import it.\n",
+    "`mpl-probscale` lets you use probability scales. All you need to do is import it.\n",
     "\n",
     "Before importing:"
    ]
@@ -143,7 +143,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "To access probability scales, simply import :py:mod:`probscale`."
+    "To access probability scales, simply import `probscale`."
    ]
   },
   {
@@ -167,11 +167,11 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "Probability scales default to the standard normal distribution (ote that the formatting is a percentage-based probability)\n",
+    "Probability scales default to the standard normal distribution (note that the formatting is a percentage-based probability)\n",
     "\n",
-    "You can even use different probability distributions, though it can be tricky. You have to pass a frozen distribution to the `dist` kwarg in `ax.set_[x|y]scale`.\n",
+    "You can even use different probability distributions, though it can be tricky. You have to pass a frozen distribution from either [scipy.stats](http://docs.scipy.org/doc/scipy/reference/stats.html#continuous-distributions) or [paramnormal](http://phobson.github.io/paramnormal/) to the `dist` kwarg in `ax.set_[x|y]scale`.\n",
     "\n",
-    "Here's a standard normal scale right next to two different beta scales for comparison"
+    "Here's a standard normal scale with two different beta scales and a linear scale for comparison."
    ]
   },
   {
@@ -182,23 +182,20 @@
    },
    "outputs": [],
    "source": [
-    "import paramnormal\n",
-    "seaborn.set(style='ticks', context='notebook', rc=clear_bkgd)\n",
-    "\n",
-    "fig, (ax1, ax2, ax3, ax4) = pyplot.subplots(figsize=(8, 6), nrows=4)\n",
+    "fig, (ax1, ax2, ax3, ax4) = pyplot.subplots(figsize=(8, 4), nrows=4)\n",
     "\n",
     "ax1.set_xscale('prob')\n",
     "ax1.set_xlim(left=2, right=98)\n",
     "ax1.set_xlabel('Normal probability scale, as percents')\n",
     "ax1.set_yticks([])\n",
     "\n",
-    "beta1 = paramnormal.beta(α=3, β=2)\n",
+    "beta1 = stats.beta(a=3, b=2)\n",
     "ax2.set_xscale('prob', dist=beta1)\n",
     "ax2.set_xlim(left=2, right=98)\n",
     "ax2.set_xlabel('Beta probability scale (α=3, β=2)')\n",
     "ax2.set_yticks([])\n",
     "\n",
-    "beta2 = paramnormal.beta(α=2, β=7)\n",
+    "beta2 = stats.beta(a=2, b=7)\n",
     "ax3.set_xscale('prob', dist=beta2)\n",
     "ax3.set_xlim(left=2, right=98)\n",
     "ax3.set_xlabel('Beta probability scale (α=2, β=7)')\n",
@@ -231,7 +228,7 @@
    "outputs": [],
    "source": [
     "numpy.random.seed(0)\n",
-    "sample = paramnormal.normal(μ=4, σ=2).rvs(size=37)\n",
+    "sample = numpy.random.normal(loc=4, scale=2, size=37)\n",
     "\n",
     "fig = probscale.probplot(sample)\n",
     "seaborn.despine(fig=fig)"
@@ -281,7 +278,7 @@
     "fig, ax = pyplot.subplots(figsize=(3, 7))\n",
     "\n",
     "numpy.random.seed(0)\n",
-    "new_sample = paramnormal.lognormal(μ=2.0, σ=0.75).rvs(size=37)\n",
+    "new_sample = numpy.random.lognormal(mean=2.0, sigma=0.75, size=37)\n",
     "\n",
     "probscale.probplot(\n",
     "    new_sample,\n",
@@ -325,11 +322,46 @@
    },
    "outputs": [],
    "source": [
-    "fig, (ax1, ax2) = pyplot.subplots(ncols=2, figsize=(7, 2))\n",
+    "fig, (ax1, ax2, ax3) = pyplot.subplots(nrows=3, figsize=(5.5, 7))\n",
+    "\n",
     "probscale.probplot(sample, ax=ax1, axtype='pp', xlabel='Percentiles')\n",
     "probscale.probplot(sample, ax=ax2, axtype='qq', xlabel='Quantiles')\n",
+    "probscale.probplot(sample, ax=ax3, axtype='prob', xlabel='Probabilities')\n",
+    "\n",
+    "ax2.set_xlim(left=-2.5, right=2.5)\n",
+    "ax3.set_xlim(left=0.5, right=99.5)\n",
+    "fig.tight_layout()\n",
     "seaborn.despine(fig=fig)"
    ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### Working with seaborn `FacetGrids`\n",
+    "\n",
+    "Good news, everyone. The ``probplot`` function generally works as expected with [FacetGrids](http://stanford.edu/~mwaskom/software/seaborn/tutorial/axis_grids.html#subsetting-data-with-facetgrid)."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {
+    "collapsed": false,
+    "scrolled": false
+   },
+   "outputs": [],
+   "source": [
+    "plot = (\n",
+    "    seaborn.load_dataset(\"tips\")\n",
+    "        .assign(pct=lambda df: 100 * df['tip'] / df['total_bill'])\n",
+    "        .pipe(seaborn.FacetGrid, hue='time', col='sex', row='smoker', margin_titles=True, aspect=1.75)\n",
+    "        .map(probscale.probplot, 'pct', bestfit=True, otherscale='log', scatter_kws=dict(alpha=0.75))\n",
+    "        .add_legend()\n",
+    "        .set_xlabels('Non-Exceedance Probabilty')\n",
+    "        .set_ylabels('Tips as percent of total bill')\n",
+    ")"
+   ]
   }
  ],
  "metadata": {
@@ -353,4 +385,4 @@
  },
  "nbformat": 4,
  "nbformat_minor": 0
-}
+}

probscale/__init__.py

@@ -4,3 +4,6 @@
 from .probscale import ProbScale
 
 scale.register_scale(ProbScale)
+
+from numpy.testing import Tester
+test = Tester().test

probscale/tests/test_probscale.py

@@ -3,7 +3,11 @@
 
 import numpy as np
 import matplotlib.pyplot as plt
-from scipy import stats
+
+try:
+    from scipy import stats
+except:
+    stats = None
 
 from matplotlib.testing.decorators import image_comparison, cleanup
 import nose.tools as nt
@@ -91,6 +95,7 @@ def test_the_scale_not_as_pct():
 
 
 @image_comparison(baseline_images=['test_the_scale_beta'], extensions=['png'])
+@nptest.dec.skipif(stats is None)
 def test_the_scale_beta():
     fig, ax = plt.subplots(figsize=(4, 8))
     ax.set_yscale('prob', as_pct=True, dist=stats.beta(3, 2))