Merge pull request #32 from phobson/add-CIs-to-fitter

Add CIs options to linear fits and bump to version 0.2.1

Author: phobson

conda.recipe/meta.yaml

@@ -1,6 +1,6 @@
 package:
   name: mpl-probscale
-  version: 0.2.0dev
+  version: 0.2.1dev
 
 source:
   path: ../

docs/conf.py

@@ -85,9 +85,9 @@
 # built documents.
 #
 # The short X.Y version.
-version = '0.2.0'
+version = '0.2.1'
 # The full version, including alpha/beta/rc tags.
-release = '0.2.0'
+release = '0.2.1'
 
 # The language for content autogenerated by Sphinx. Refer to documentation
 # for a list of supported languages.

docs/tutorial/closer_look_at_viz.ipynb

@@ -499,15 +499,10 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## Controlling the aesthetics of the plot elements\n",
-    "\n",
-    "As it has been hinted in the examples above, the `probplot` function takes two dictionaries to customize the data series and the best-fit line (`scatter_kws` and `line_kws`, respectively.\n",
-    "These dictionaries are passed directly to the `plot` method of current axes.\n",
+    "### Bootstrapped confidence intervals\n",
     "\n",
-    "By default, the data series assumes that `linestyle='none'` and `marker='o'`.\n",
-    "These can be overwritten through `scatter_kws`\n",
-    "\n",
-    "Revisting the previous example, we can customize it like so:"
+    "Regardless of the scales of the plot (linear, log, or prob), you can add bootstrapped confidence intervals around the best-fit line. \n",
+    "Simply use the ``estimate_ci=True`` option along with ``bestfit=True``:"
    ]
   },
   {
@@ -518,45 +513,16 @@
    },
    "outputs": [],
    "source": [
-    "scatter_options = dict(\n",
-    "    marker='^',\n",
-    "    markerfacecolor='none',\n",
-    "    markeredgecolor='firebrick',\n",
-    "    markeredgewidth=1.25,\n",
-    "    linestyle='none',\n",
-    "    alpha=0.35,\n",
-    "    zorder=5,\n",
-    "    label='Meal Cost ($)'\n",
-    ")\n",
-    "\n",
-    "line_options = dict(\n",
-    "    dashes=(10,2,5,2,10,2),\n",
-    "    color='0.25',\n",
-    "    linewidth=3,\n",
-    "    zorder=10,\n",
-    "    label='Best-fit line'\n",
-    ") \n",
-    "\n",
-    "fig, ax = pyplot.subplots(figsize=(4, 6))\n",
-    "fig = probscale.probplot(tips['total_bill'], ax=ax, plottype='prob', probax='y', bestfit=True, \n",
-    "                         datascale='log', problabel='Probabilities', datalabel='Total Bill (USD)',\n",
-    "                         scatter_kws=scatter_options, line_kws=line_options)\n",
-    "ax.legend(loc='upper left')\n",
-    "ax.set_ylim(bottom=0.1, top=99.9)\n",
-    "seaborn.despine()"
-   ]
-  },
-  {
-   "cell_type": "raw",
-   "metadata": {},
-   "source": [
-    ".. note::\n",
-    "   The ``probplot`` function can take two additional aesthetic parameters:\n",
-    "   `color` and `label`. If provided, `color` will override the marker face color\n",
-    "   and line color options of the `scatter_kws` and `line_kws` parameters, respectively.\n",
-    "   Similarly, the label of the scatter series will be overridden by the explicit parameter.\n",
-    "   It is not recommended that `color` and `label` are used. They exist primarily for \n",
-    "   compatibility with the seaborn package."
+    "N = 15\n",
+    "numpy.random.seed(0)\n",
+    "x = numpy.random.normal(size=N) + numpy.random.uniform(size=N)\n",
+    "fig, ax = pyplot.subplots(figsize=(8, 4))\n",
+    "fig = probscale.probplot(x, ax=ax, bestfit=True, estimate_ci=True, \n",
+    "                         line_kws={'label': 'BF Line', 'color': 'b'},\n",
+    "                         scatter_kws={'label': 'Observations'},\n",
+    "                         problabel='Probability (%)')\n",
+    "ax.legend(loc='lower right')\n",
+    "seaborn.despine(fig)"
    ]
   },
   {
@@ -575,37 +541,10 @@
     "You should read that function's docstring for more detailed information.\n",
     "But the high-level overview is that there are a couple of parameters (``alpha`` and ``beta``) that you can tweak in the plotting positions calculation.\n",
     "\n",
-    "The most common values can be selected via the ``postype`` parameter and are described as follows:\n",
-    "\n",
-    "    \"type 4\" (α=0, β=1)\n",
-    "        Linear interpolation of the empirical CDF.\n",
-    "    \"type 5\" or \"hazen\" (α=0.5, β=0.5)\n",
-    "        Piecewise linear interpolation.\n",
-    "    \"type 6\" or \"weibull\" (α=0, β=0)\n",
-    "        Weibull plotting positions. Unbiased exceedance probability\n",
-    "        for all distributions. Recommended for hydrologic applications.\n",
-    "    \"type 7\" (α=1, β=1)\n",
-    "        The default values in R. Not recommended with probability\n",
-    "        scales as the min and max data points get plotting positions\n",
-    "        of 0 and 1, respectively, and therefore cannot be shown.\n",
-    "    \"type 8\" (α=1/3, β=1/3)\n",
-    "        Approximately median-unbiased.\n",
-    "    \"type 9\" or \"blom\" (α=0.375, β=0.375)\n",
-    "        Approximately unbiased positions if the data are normally\n",
-    "        distributed.\n",
-    "    \"median\" (α=0.3175, β=0.3175)\n",
-    "        Median exceedance probabilities for all distributions\n",
-    "        (used in ``scipy.stats.probplot``).\n",
-    "    \"apl\" or \"pwm\" (α=0.35, β=0.35)\n",
-    "        Used with probability-weighted moments.\n",
-    "    \"cunnane\" (α=0.4, β=0.4)\n",
-    "        Nearly unbiased quantiles for normally distributed data.\n",
-    "        This is the default value.\n",
-    "    \"gringorten\" (α=0.44, β=0.44)\n",
-    "        Used for Gumble distributions.\n",
-    "        \n",
+    "The most common values can be selected via the ``postype`` parameter.\n",
+    "\n",
     "These are controlled via the `pp_kws` parameter in `probplot`\n",
-    "and are discussed in much more detail in the next tutorial.\n"
+    "and are discussed in much more detail in the `next tutorial <closer_look_at_plot_pos.html>`_."
    ]
   },
   {
@@ -639,6 +578,70 @@
     "fig.tight_layout()"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Controlling the aesthetics of the plot elements\n",
+    "\n",
+    "As it has been hinted in the examples above, the `probplot` function takes two dictionaries to customize the data series and the best-fit line (`scatter_kws` and `line_kws`, respectively.\n",
+    "These dictionaries are passed directly to the `plot` method of current axes.\n",
+    "\n",
+    "By default, the data series assumes that `linestyle='none'` and `marker='o'`.\n",
+    "These can be overwritten through `scatter_kws`\n",
+    "\n",
+    "Revisting the previous example, we can customize it like so:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {
+    "collapsed": false
+   },
+   "outputs": [],
+   "source": [
+    "scatter_options = dict(\n",
+    "    marker='^',\n",
+    "    markerfacecolor='none',\n",
+    "    markeredgecolor='firebrick',\n",
+    "    markeredgewidth=1.25,\n",
+    "    linestyle='none',\n",
+    "    alpha=0.35,\n",
+    "    zorder=5,\n",
+    "    label='Meal Cost ($)'\n",
+    ")\n",
+    "\n",
+    "line_options = dict(\n",
+    "    dashes=(10,2,5,2,10,2),\n",
+    "    color='0.25',\n",
+    "    linewidth=3,\n",
+    "    zorder=10,\n",
+    "    label='Best-fit line'\n",
+    ") \n",
+    "\n",
+    "fig, ax = pyplot.subplots(figsize=(4, 6))\n",
+    "fig = probscale.probplot(tips['total_bill'], ax=ax, plottype='prob', probax='y', bestfit=True, \n",
+    "                         datascale='log', problabel='Probabilities', datalabel='Total Bill (USD)',\n",
+    "                         scatter_kws=scatter_options, line_kws=line_options)\n",
+    "ax.legend(loc='upper left')\n",
+    "ax.set_ylim(bottom=0.1, top=99.9)\n",
+    "seaborn.despine()"
+   ]
+  },
+  {
+   "cell_type": "raw",
+   "metadata": {},
+   "source": [
+    ".. note::\n",
+    "   The ``probplot`` function can take two additional aesthetic parameters:\n",
+    "   `color` and `label`. If provided, `color` will override the marker face color\n",
+    "   and line color options of the `scatter_kws` and `line_kws` parameters, respectively.\n",
+    "   Similarly, the label of the scatter series will be overridden by the explicit parameter.\n",
+    "   It is not recommended that `color` and `label` are used. They exist primarily for \n",
+    "   compatibility with the seaborn package."
+   ]
+  },
   {
    "cell_type": "markdown",
    "metadata": {},
@@ -716,7 +719,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.5.1"
+   "version": "3.5.2"
   }
  },
  "nbformat": 4,

docs/tutorial/getting_started.ipynb

@@ -294,6 +294,7 @@
     "    probax='y', # flip the plot\n",
     "    datascale='log', # scale of the non-probability axis\n",
     "    bestfit=True, # draw a best-fit line\n",
+    "    estimate_ci=True,\n",
     "    datalabel='Lognormal Values',  # labels and markers...\n",
     "    problabel='Non-exceedance probability',\n",
     "    scatter_kws=dict(marker='d', zorder=2, mew=1.25, mec='w', markersize=10),\n",
@@ -388,7 +389,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.5.1"
+   "version": "3.5.2"
   }
  },
  "nbformat": 4,