Basic summary stat plotting

examples/plot_adjacency_matrix.py

@@ -0,0 +1,24 @@
+"""
+Plot an Adjacency Matrix
+---------------------
+
+"""
+
+from grave.stats import plot_adjacency_matrix
+from networkx.generators.random_graphs import barabasi_albert_graph
+import matplotlib.pyplot as plt
+
+# Generating a networkx graph
+graph = barabasi_albert_graph(50, 3)
+
+fig, ax_mat = plt.subplots(figsize=(12, 6), ncols=2)
+plot_adjacency_matrix(graph, ax=ax_mat[0])
+ax_mat[0].set_title('Default Style', x=0.5, y=-0.1)
+
+plot_adjacency_matrix(graph,
+                      xticklabels=False,
+                      yticklabels=False,
+                      linewidths=0,
+                      ax=ax_mat[1])
+ax_mat[1].set_title('A Minimalist Style', x=0.5, y=-0.1)
+plt.show()

examples/plot_adjacency_matrix_labelfunc.py

@@ -0,0 +1,20 @@
+"""
+Plot an Adjacency Matrix with Custom Labels
+---------------------
+
+"""
+
+from grave.stats import plot_adjacency_matrix
+from networkx.generators.random_graphs import barabasi_albert_graph
+import matplotlib.pyplot as plt
+
+# Generating a networkx graph
+graph = barabasi_albert_graph(50, 3)
+
+for node, node_attrs in graph.nodes.data():
+    node_attrs['label'] = 'Node {0}'.format(str(node))
+
+
+fig, ax= plt.subplots(figsize=(8, 8))
+plot_adjacency_matrix(graph, ax=ax)
+plt.show()

examples/plot_weighted_adjacency.py

@@ -0,0 +1,30 @@
+"""
+Plot a Weighted Adjacency Matrix
+---------------------
+
+Basic example of weighted adjacency matrix plotting.
+
+"""
+
+from grave.stats import plot_adjacency_matrix
+from grave import plot_network
+from networkx.generators.random_graphs import barabasi_albert_graph
+import matplotlib.pyplot as plt
+import numpy as np
+
+# Generating a networkx graph
+network = barabasi_albert_graph(50, 3)
+
+# Give it random edge weights
+weights = np.random.normal(loc=10, scale=5, size=network.number_of_edges())
+for w, (u, v, attrs) in zip(weights, network.edges.data()):
+    attrs['weight'] = w
+
+fig, ax_mat = plt.subplots(figsize=(16, 8), ncols=2)
+
+plot_network(network, ax=ax_mat[0])
+
+plot_adjacency_matrix(network,
+                      weighted=True,
+                      ax=ax_mat[1])
+plt.show()

grave/stats.py

@@ -0,0 +1,152 @@
+import warnings
+import functools
+
+import networkx as nx
+import numpy as np
+import matplotlib.cm as cm
+import matplotlib.colors as mcolors
+
+from .grave import _ensure_ax
+
+
+def _optional_dependency(dependency):
+    def _optional_dependency(func):
+        @functools.wraps(func)
+        def wrapper(*args, **kwargs):
+            try:
+                return func(*args, **kwargs)
+            except ImportError as e:
+                if e.name == dependency:
+                    warnings.warn('Optional dependency {0} not installed, '
+                                  'returning None!'.format(dependency))
+                    return None
+                else:
+                    raise e
+        return wrapper
+    return _optional_dependency
+
+
+@_optional_dependency('seaborn')
+@_ensure_ax
+def plot_adjacency_matrix(network,
+                          node_labels=None,
+                          label_behavior=None,
+                          weighted=False,
+                          frame=True,
+                          xtickrotation=70,
+                          ytickrotation=0,
+                          *, ax,
+                          **heatmap_kwargs):
+    '''Plot the adjacency matrix of a network. If weight is True, 
+    use a `weight` attribute from the edges to plot a heatmap of weights.
+
+    Requires seaborn to be installed.
+    Extra keyword parameters are passed on to seaborn's `heatmap` function.
+
+    Parameters
+    ----------
+
+    network : networkx graph object
+    node_labels : callable, "auto", int, or iterable, optional
+        If callable, should be a function taking a node attribute dict
+        and returning a string. If None, checks each node for a label
+        attribute and uses it if found, or uses str(node).
+    weighted : bool, optional
+        If True, draw a weighted adjacency matrix using a `weight`
+        edge attribute.
+    frame : bool, optional
+        If True, draw a wider frame around the matrix.
+    xtickrotation : int, optional
+        Rotation to apply to x axis labels.
+    ytickrotation : int, optional
+        Rotation to apply to y axis labels.
+
+    Returns
+    -------
+    The matplotlib axes.
+    '''
+    from seaborn import heatmap
+    import pandas as pd
+    import matplotlib.pyplot as plt
+    import warnings
+
+    params = {'vmin': None,
+              'vmax': None,
+              'cmap': None,
+              'center': None,
+              'robust': False,
+              'annot': None,
+              'fmt': '.2g',
+              'annot_kws': None,
+              'linecolor': 'lightgray',
+              'linewidths': .5,
+              'cbar': False,
+              'cbar_kws': {'shrink': .5},
+              'cbar_ax': None,
+              'xticklabels': 'auto',
+              'yticklabels': 'auto',
+              'square': True,
+              'mask': None}
+
+    if weighted:
+        adj_mat = np.empty((network.number_of_nodes(),
+                            network.number_of_nodes()))
+        adj_mat[:] = np.NaN
+
+        missing_weight = 0
+        directed = nx.is_directed(network)
+        node_idx = {node : idx for idx, node in \
+                    enumerate(network.nodes.keys())}
+        for u, v, edge_attrs in network.edges.data():
+            u_idx = node_idx[u]
+            v_idx = node_idx[v]
+            try:
+                weight = edge_attrs['weight']
+            except KeyError:
+                weight = 0
+                missing_weight += 1
+            adj_mat[u_idx, v_idx] = weight
+            if not directed:
+                adj_mat[v_idx, u_idx] = weight
+
+        params['cbar'] = True
+        params['cmap'] = plt.get_cmap()
+
+        if missing_weight > 0:
+            n_edges = network.number_of_edges()
+            warnings.warn('{missing} of {n_edges}'
+                          ' edges missing weight attr,'
+                          ' using 0 for them.'.format(missing=missing_weight,
+                                                      n_edges=n_edges))
+    else:
+        adj_mat = nx.adj_matrix(network).todense()
+        cmap = plt.get_cmap('binary')
+        params['cmap'] = cmap
+
+    labels = []
+    if callable(node_labels):
+        for item, item_attr in network.nodes.data():
+            attrs = dict(item_attr)
+            labels.append(node_labels(item_attr))
+    else:
+        for node, node_attr in network.nodes.data():
+            labels.append(node_attr.get('label', str(node)))
+
+    data = pd.DataFrame(adj_mat, columns=labels, index=labels)
+
+    params.update(heatmap_kwargs)
+    ax = heatmap(data, ax=ax, **params)
+
+    if frame:
+        for axis in ['top','bottom','left','right']:
+            ax.spines[axis].set_visible(True)
+            ax.spines[axis].set_color(params['linecolor'])
+            ax.spines[axis].set_linewidth(2 * params['linewidths'])
+
+    ax.xaxis.tick_top()
+    for tick in ax.get_xticklabels():
+        tick.set_rotation(xtickrotation)
+    for tick in ax.get_yticklabels():
+        tick.set_rotation(ytickrotation)
+
+    return ax, data