Merge pull request #3 from prmiles/update_coverage

Update coverage

Author: Paul Miles

mcmcplot/mcmatplot.py

@@ -196,11 +196,11 @@ def plot_pairwise_correlation_panel(chains, names = None, settings = None):
     'skip': 1,
     'fig': dict(figsize = (7,5), dpi = 100),
     'plot': dict(color = 'b', marker = '.', linestyle = 'none'),
-    'xlabel': {'s': 'Iteration'},
+    'xlabel': {},
     'ylabel': {},
     'title': {},
     'add_5095_contours': False,
-    'plot_50': dict(color = 'r', marker = None, linewidth = 2, linestyle = '--', label = '50%'),
+    'plot_50': dict(color = 'g', marker = None, linewidth = 2, linestyle = '--', label = '50%'),
     'plot_95': dict(color = 'r', marker = None, linewidth = 2, linestyle = '--', label = '95%'),
     'add_legend': False,
     'legend': dict(loc = 1),
@@ -300,7 +300,11 @@ def plot_chain_metrics(chain, name = None, settings = None):
 
 class Plot:
     '''
-    Plotting routines for analyzing sampling chains from MCMC process.
+    Wrapper routines for analyzing/plotting sampling chains from MCMC process.
+    
+    Uses methods from the `matplotlib` package:
+    
+    https://matplotlib.org/
 
     Attributes:
         - :meth:`~plot_density_panel`

mcmcplot/mcseaborn.py

@@ -29,7 +29,7 @@ def plot_joint_distributions(chains, names = None, sns_style = 'white', settings
         'skip': 1,
         'sns_style': sns_style,
         'sns': sns.axes_style(style = sns_style),
-        'jointplot': dict(kind='kde', height=6, space=0)
+        'jointplot': dict(kind='kde', data = None, height = 6.0, space=0)
         }
 
     settings = check_settings(default_settings = default_settings, user_settings = settings)
@@ -46,7 +46,8 @@ def plot_joint_distributions(chains, names = None, sns_style = 'white', settings
             chain2 = pd.Series(chains[inds,jj-1], name=names[jj-1])
 
             # Show the joint distribution using kernel density estimation
-            g.append(sns.jointplot(chain1, chain2, **settings['jointplot']))
+            a = sns.jointplot(x = chain1, y = chain2, **settings['jointplot'])
+            g.append(a)
 
     return g, settings
 
@@ -91,4 +92,20 @@ def plot_paired_density_matrix(chains, names = None, sns_style = 'white', index
     g.map_lower(settings['ld_type'], **settings['ld'])
     g.map_upper(settings['ud_type'], **settings['ud'])
     g.map_diag(settings['md_type'], **settings['md'])
-    return g, settings
+    return g, settings
+
+class Plot:
+    '''
+    Wrapper routines for analyzing/plotting sampling chains from MCMC process.
+    
+    Uses methods from the `seaborn` package:
+    
+    https://seaborn.pydata.org/
+
+    Attributes:
+        - :meth:`~plot_joint_distributions`
+        - :meth:`~plot_paired_density_matrix`
+    '''
+    def __init__(self):
+        self.plot_joint_distributions = plot_joint_distributions
+        self.plot_paired_density_matrix = plot_paired_density_matrix

mcmcplot/utilities.py

@@ -6,43 +6,44 @@
 @author: prmiles
 """
 import numpy as np
-from scipy.interpolate import interp1d
 from scipy import pi,sin,cos
 import sys
 import math
 
 def check_settings(default_settings, user_settings = None):
-        
-        settings = default_settings.copy()
-        
-        options = list(default_settings.keys())
-        if user_settings is None:
-            user_settings = {}
-        user_options = list(user_settings.keys())
-        for ii in range(len(user_options)):
-            if user_options[ii] in options:
-                # check if checking a dictionary
-                if isinstance(settings[user_options[ii]], dict):
-                    settings[user_options[ii]] = check_settings(settings[user_options[ii]], user_settings[user_options[ii]])
-                else:
-                    settings[user_options[ii]] = user_settings[user_options[ii]]
-            if user_options[ii] not in options:
-                settings[user_options[ii]] = user_settings[user_options[ii]]
-                
-        return settings
+    '''
+    Check user settings with default.
+    
+    Recursively checks elements of user settings against the defaults and updates settings
+    as it goes.  If a user setting does not exist in the default, then the user setting
+    is added to the settings.  If the setting is defined in both the user and default
+    settings, then the user setting overrides the default.  Otherwise, the default
+    settings persist.
     
-def generate_plotly_subplot_coords(nparam, ns1, ns2):
-    sprow = []
-    spcol = []
-    counter = 0
-    for ii in range(ns1):
-        for jj in range(ns2):
-            sprow.append(ii + 1)
-            spcol.append(jj + 1)
-            counter += 1
-            if counter is nparam:
-                break
-    return sprow, spcol
+    Args:
+        * **default_settings** (:py:class:`dict`): Default settings for particular method.
+        * **user_settings** (:py:class:`dict`): User defined settings.
+    
+    Returns:
+        * (:py:class:`dict`): Updated settings.
+    '''
+    settings = default_settings.copy() # initially define settings as default
+    
+    options = list(default_settings.keys()) # get default settings
+    if user_settings is None: # convert to empty dict
+        user_settings = {}
+    user_options = list(user_settings.keys()) # get user settings
+    for uo in user_options: # iterate through settings
+        if uo in options:
+            # check if checking a dictionary
+            if isinstance(settings[uo], dict):
+                settings[uo] = check_settings(settings[uo], user_settings[uo])
+            else:
+                settings[uo] = user_settings[uo]
+        if uo not in options:
+            settings[uo] = user_settings[uo]
+            
+    return settings
 
 def generate_subplot_grid(nparam = 2):
     '''
@@ -85,30 +86,6 @@ def generate_names(nparam, names):
         names = extend_names_to_match_nparam(names, nparam)
     return names
 
-def setup_plot_features(nparam, names, figsizeinches):
-    '''
-    Setup plot features.
-
-    Args:
-        * **nparam** (:py:class:`int`): Number of parameters
-        * **names** (:py:class:`list`): Names of parameters provided by user
-        * **figsizeinches** (:py:class:`list`): [Width, Height]
-
-    Returns:
-        * **ns1** (:py:class:`int`): Number of rows in subplot
-        * **ns2** (:py:class:`int`): Number of columns in subplot
-        * **names** (:py:class:`list`): List of strings - parameter names
-        * **figsizeiches** (:py:class:`list`): [Width, Height]
-    '''
-    ns1, ns2 = generate_subplot_grid(nparam = nparam)
-
-    names = generate_names(nparam = nparam, names = names)
-    
-    if figsizeinches is None:
-        figsizeinches = [5,4]
-        
-    return ns1, ns2, names, figsizeinches
-
 def generate_default_names(nparam):
     '''
     Generate generic parameter name set.
@@ -348,86 +325,4 @@ def append_to_nrow_ncol_based_on_shape(sh, nrow, ncol):
     else:
         nrow.append(sh[0])
         ncol.append(sh[1])
-    return nrow, ncol
-
-# --------------------------------------------
-def convert_flag_to_boolean(flag):
-    '''
-    Convert flag to boolean for backwards compatibility.
-
-    Args:
-        * **flag** (:py:class:`bool` or :py:class:`int`): Flag to specify something.
-
-    Returns:
-        * **flag** (:py:class:`bool`): Flag to converted to boolean.
-    '''
-    if flag is 'on':
-        flag = True
-    elif flag is 'off':
-        flag = False
-        
-    return flag
-
-# --------------------------------------------
-def set_local_parameters(ii, local):
-    '''
-    Set local parameters based on tests.
-    
-    :Test 1:
-        * `local == 0`
-    :Test 2:
-        * `local == ii`
-    
-    Args:
-        * **ii** (:py:class:`int`): Index.
-        * **local** (:class:`~numpy.ndarray`): Local flags.
-
-    Returns:
-        * **test** (:class:`~numpy.ndarray`): Array of Booleans indicated test results.
-    '''
-    # some parameters may only apply to certain batch sets
-    test1 = local == 0
-    test2 = local == ii
-    test = test1 + test2
-    return test.reshape(test.size,)
-
-# --------------------------------------------
-def empirical_quantiles(x, p = np.array([0.25, 0.5, 0.75])):
-    '''
-    Calculate empirical quantiles.
-
-    Args:
-        * **x** (:class:`~numpy.ndarray`): Observations from which to generate quantile.
-        * **p** (:class:`~numpy.ndarray`): Quantile limits.
-
-    Returns:
-        * (:class:`~numpy.ndarray`): Interpolated quantiles.
-    '''
-
-    # extract number of rows/cols from np.array
-    n = x.shape[0]
-    # define vector valued interpolation function
-    xpoints = range(n)
-    interpfun = interp1d(xpoints, np.sort(x, 0), axis = 0)
-    
-    # evaluation points
-    itpoints = (n-1)*p
-    
-    return interpfun(itpoints)
-
-# --------------------------------------------    
-def check_defaults(kwargs, defaults):
-    '''
-    Check if defaults are defined in kwargs
-    
-    Args:
-        * **kwargs** (:py:class:`dict`): Keyword arguments.
-        * **defaults** (:py:class:`dict`): Default settings.
-
-    Returns:
-        * **kwargs** (:py:class:`dict`): Updated keyword arguments with at least defaults set.
-    '''
-    for ii in defaults:
-        if ii not in kwargs:
-            kwargs[ii] = defaults[ii]
-    return kwargs
+    return nrow, ncol

test/test_mcmatplot.py

@@ -12,7 +12,7 @@
 import math
 
 import unittest
-
+            
 # --------------------------
 class PlotDensityPanel(unittest.TestCase):
     def test_basic_plot_features(self):
@@ -66,6 +66,21 @@ def test_basic_plot_features_nsimu_gt_maxpoints(self):
         self.assertEqual(f.axes[1].get_xlabel(), 'Iteration', msg = 'Should be Iteration')
         plt.close()
 
+    def test_basic_plot_features_nsimu_gt_maxpoints_with_pm2std(self):
+        nsimu = 1000
+        chains = np.random.random_sample(size = (nsimu,2))
+        f, _ = MP.plot_chain_panel(chains = chains, settings = dict(add_pm2std = True))
+        x1, y1 = f.axes[0].lines[0].get_xydata().T
+        x2, y2 = f.axes[1].lines[0].get_xydata().T
+        skip = int(math.floor(nsimu/500))
+        self.assertTrue(np.array_equal(y1, chains[range(0,nsimu,skip),0]), msg = 'Expect y1 to match column 1')
+        self.assertTrue(np.array_equal(y2, chains[range(0,nsimu,skip),1]), msg = 'Expect y2 to match column 2')
+        self.assertEqual(f.axes[0].get_xlabel(), '', msg = 'Should be blank')
+        self.assertEqual(f.axes[1].get_xlabel(), 'Iteration', msg = 'Should be Iteration')
+        self.assertEqual(len(f.axes[0].lines), 4, msg = 'Expect 4 lines')
+        self.assertEqual(len(f.axes[1].lines), 4, msg = 'Expect 4 lines')
+        plt.close()
+        
 # --------------------------
 class PlotHistogramPanel(unittest.TestCase):
     def test_basic_plot_features_nsimu_lt_maxpoints(self):
@@ -106,6 +121,43 @@ def test_basic_plot_features_nsimu_lt_maxpoints(self):
 
         plt.close()
 
+    def test_basic_plot_features_nsimu_lt_maxpoints_and_2_chains(self):
+        chains = np.random.random_sample(size = (100,2))
+        f, _ = MP.plot_pairwise_correlation_panel(chains = chains)
+        x1, y1 = f.axes[0].lines[0].get_xydata().T
+        self.assertTrue(np.array_equal(x1, chains[:,0]), msg = 'Expect x1 to match column 0')
+        self.assertTrue(np.array_equal(y1, chains[:,1]), msg = 'Expect y1 to match column 1')
+        for ai in f.axes:
+            self.assertEqual(ai.get_title(), '', msg = 'Should be blank')
+        self.assertEqual(f.axes[0].get_xlabel(),'$p_{0}$', msg = 'Expect $p_{0}$')
+        self.assertEqual(f.axes[0].get_ylabel(),'$p_{1}$', msg = 'Expect $p_{1}$')
+        plt.close()
+        
+    def test_basic_plot_features_2c_w_contours(self):
+        chains = np.random.random_sample(size = (100,2))
+        f, _ = MP.plot_pairwise_correlation_panel(chains = chains, settings = dict(add_5095_contours = True))
+        x1, y1 = f.axes[0].lines[0].get_xydata().T
+        
+        self.assertTrue(np.array_equal(x1, chains[:,0]), msg = 'Expect x1 to match column 0')
+        self.assertTrue(np.array_equal(y1, chains[:,1]), msg = 'Expect y1 to match column 1')
+        self.assertEqual(len(f.axes[0].lines), 3, msg = 'Expect 3 lines')
+        for ai in f.axes:
+            self.assertEqual(ai.get_title(), '', msg = 'Should be blank')
+        self.assertEqual(f.axes[0].get_xlabel(),'$p_{0}$', msg = 'Expect $p_{0}$')
+        self.assertEqual(f.axes[0].get_ylabel(),'$p_{1}$', msg = 'Expect $p_{1}$')
+        plt.close()
+        
+    def test_basic_plot_features_2c_w_contours_and_legend(self):
+        chains = np.random.random_sample(size = (100,2))
+        f, _ = MP.plot_pairwise_correlation_panel(chains = chains, settings = dict(add_5095_contours = True, add_legend = True))
+        x1, y1 = f.axes[0].lines[0].get_xydata().T
+        
+        self.assertTrue(np.array_equal(x1, chains[:,0]), msg = 'Expect x1 to match column 0')
+        self.assertTrue(np.array_equal(y1, chains[:,1]), msg = 'Expect y1 to match column 1')
+        self.assertEqual(len(f.axes[0].lines), 3, msg = 'Expect 3 lines')
+        self.assertEqual(len(f.legends), 1, msg = 'Expect legend')
+        plt.close()
+        
 # --------------------------
 class PlotChainMetrics(unittest.TestCase):
     def test_basic_plot_features(self):

test/test_mcseaborn.py

@@ -0,0 +1,49 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""
+Created on Thu Jun 21 12:21:24 2018
+
+@author: prmiles
+"""
+
+from mcmcplot import mcseaborn as MP
+from mcmcplot import utilities
+import matplotlib.pyplot as plt
+import seaborn as sns
+import numpy as np
+
+import unittest
+            
+# --------------------------
+class Plot(unittest.TestCase):
+    def test_plot_attributes(self):
+        P = MP.Plot()
+        check_these = ['plot_joint_distributions', 'plot_paired_density_matrix']
+        for ct in check_these:
+            self.assertTrue(hasattr(P, ct), msg = str('P contains method {}'.format(ct)))
+            
+# --------------------------
+class PlotJointDistributions(unittest.TestCase):
+    def test_basic_joint_distributions(self):
+        npar = 3
+        chains = np.random.random_sample(size = (100,npar))
+        f, _ = MP.plot_joint_distributions(chains = chains)
+        names = utilities.generate_names(nparam = npar, names = None)
+        count = 0
+        for jj in range(2,npar+1):
+            for ii in range(1,jj):
+                name1 = names[ii-1]
+                name2 = names[jj-1]
+                self.assertEqual(f[count].ax_joint.get_xlabel(), name1, msg = str('Should be {}'.format(name1)))
+                self.assertEqual(f[count].ax_joint.get_ylabel(), name2, msg = str('Should be {}'.format(name2)))
+                count += 1
+        plt.close()
+              
+# --------------------------
+class PlotPairedGrid(unittest.TestCase):
+    def test_basic_paired_grid(self):
+        npar = 3
+        chains = np.random.random_sample(size = (100,npar))
+        f, _ = MP.plot_paired_density_matrix(chains = chains)
+        self.assertTrue(isinstance(f, sns.axisgrid.PairGrid), msg = 'Expect seaborn axisgrid PairGrid object')
+        plt.close()