Merge pull request #213 from andrewheusser/remove-hdbscan-dep

Remove hdbscan dependency

Author: jeremymanning

examples/animate.py

@@ -18,4 +18,4 @@
 geo = hyp.load('weights_avg')
 
 # plot
-geo.plot(animate=True)
+geo.plot(animate=True, legend=['first', 'second'])

hypertools/tools/cluster.py

@@ -1,12 +1,28 @@
 #!/usr/bin/env python
 import warnings
-from sklearn.cluster import KMeans, MiniBatchKMeans, AgglomerativeClustering, Birch, FeatureAgglomeration, SpectralClustering
-import numpy as np
 import six
-from hdbscan import HDBSCAN
+import numpy as np
+from sklearn.cluster import KMeans, MiniBatchKMeans, AgglomerativeClustering, Birch, FeatureAgglomeration, SpectralClustering
 from .._shared.helpers import *
 from .format_data import format_data as formatter
 
+# dictionary of models
+models = {
+    'KMeans': KMeans,
+    'MiniBatchKMeans': MiniBatchKMeans,
+    'AgglomerativeClustering': AgglomerativeClustering,
+    'FeatureAgglomeration': FeatureAgglomeration,
+    'Birch': Birch,
+    'SpectralClustering': SpectralClustering,
+}
+
+try:
+    from hdbscan import HDBSCAN
+    _has_hdbscan = True
+    models.update({'HDBSCAN': HDBSCAN})
+except ImportError:
+    _has_hdbscan = False
+
 
 @memoize
 def cluster(x, cluster='KMeans', n_clusters=3, ndims=None, format_data=True):
@@ -46,48 +62,39 @@ def cluster(x, cluster='KMeans', n_clusters=3, ndims=None, format_data=True):
 
     """
 
-    # if cluster is None, just return data
-    if cluster is None:
+    if cluster == None:
         return x
-    else:
-
-        if ndims is not None:
-            warnings.warn('The ndims argument is now deprecated. Ignoring dimensionality reduction step.')
-
-        if format_data:
-            x = formatter(x, ppca=True)
-
-        # dictionary of models
-        models = {
-            'KMeans' : KMeans,
-            'MiniBatchKMeans' : MiniBatchKMeans,
-            'AgglomerativeClustering' : AgglomerativeClustering,
-            'FeatureAgglomeration' : FeatureAgglomeration,
-            'Birch' : Birch,
-            'SpectralClustering' : SpectralClustering,
-            'HDBSCAN' : HDBSCAN
-        }
-
-        # if reduce is a string, find the corresponding model
-        if isinstance(cluster, six.string_types):
-            model = models[cluster]
-            if cluster != 'HDBSCAN':
-                model_params = {
-                    'n_clusters' : n_clusters
-                }
-            else:
-                model_params = {}
-        # if its a dict, use custom params
-        elif type(cluster) is dict:
-            if isinstance(cluster['model'], six.string_types):
-                model = models[cluster['model']]
-                model_params = cluster['params']
-
-        # initialize model
-        model = model(**model_params)
-
-        # fit the model
-        model.fit(np.vstack(x))
-
-        # return the labels
-        return list(model.labels_)
+    elif (isinstance(cluster, six.string_types) and cluster=='HDBSCAN') or \
+    (isinstance(cluster, dict) and cluster['model']=='HDBSCAN'):
+        if not _has_hdbscan:
+            raise ImportError('HDBSCAN is not installed. Please install hdbscan>=0.8.11')
+
+    if ndims != None:
+        warnings.warn('The ndims argument is now deprecated. Ignoring dimensionality reduction step.')
+
+    if format_data:
+        x = formatter(x, ppca=True)
+
+    # if reduce is a string, find the corresponding model
+    if isinstance(cluster, six.string_types):
+        model = models[cluster]
+        if cluster != 'HDBSCAN':
+            model_params = {
+                'n_clusters' : n_clusters
+            }
+        else:
+            model_params = {}
+    # if its a dict, use custom params
+    elif type(cluster) is dict:
+        if isinstance(cluster['model'], six.string_types):
+            model = models[cluster['model']]
+            model_params = cluster['params']
+
+    # initialize model
+    model = model(**model_params)
+
+    # fit the model
+    model.fit(np.vstack(x))
+
+    # return the labels
+    return list(model.labels_)

requirements.txt

@@ -6,7 +6,6 @@ matplotlib>=1.5.1
 scipy>=1.0.0
 numpy>=1.10.4
 umap-learn>=0.1.5
-hdbscan>=0.8.11
 future
 requests
 deepdish

setup.py

@@ -25,7 +25,7 @@
 
 setup(
     name='hypertools',
-    version='0.5.0',
+    version='0.5.1',
     description=DESCRIPTION,
     long_description=LONG_DESCRIPTION,
     author='Contextual Dynamics Lab',
@@ -42,7 +42,6 @@
    'matplotlib>=1.5.1',
    'scipy>=1.0.0',
    'numpy>=1.10.4',
-   'hdbscan>=0.8.11',
    'umap-learn>=0.1.5',
    'future',
    'requests',

tests/test_cluster.py

@@ -1,13 +1,14 @@
 # -*- coding: utf-8 -*-
 
 import numpy as np
+import pytest
 from hypertools.tools.cluster import cluster
 from hypertools.plot.plot import plot
 
 cluster1 = np.random.multivariate_normal(np.zeros(3), np.eye(3), size=100)
 cluster2 = np.random.multivariate_normal(np.zeros(3)+100, np.eye(3), size=100)
-data = np.vstack([cluster1,cluster2])
-labels = cluster(data,n_clusters=2)
+data = np.vstack([cluster1, cluster2])
+labels = cluster(data, n_clusters=2)
 
 
 def test_cluster_n_clusters():
@@ -19,12 +20,15 @@ def test_cluster_returns_list():
 
 
 def test_cluster_hdbscan():
-    # Given well separated clusters this should "just work"
-    hdbscan_labels = cluster(data, cluster='HDBSCAN')
-    assert len(set(hdbscan_labels)) == 2
-
-
-def text_cluster_geo():
-    geo = plot(data, show=False)
-    hdbscan_labels = cluster(geo, cluster='HDBSCAN')
-    assert len(set(hdbscan_labels)) == 2
+    try:
+        from hdbscan import HDBSCAN
+        _has_hdbscan = True
+    except:
+        _has_hdbscan = False
+
+    if _has_hdbscan:
+        hdbscan_labels = cluster(data, cluster='HDBSCAN')
+        assert len(set(hdbscan_labels)) == 2
+    else:
+        with pytest.raises(ImportError):
+            hdbscan_labels = cluster(data, cluster='HDBSCAN')

tests/test_plot.py

@@ -99,13 +99,6 @@ def test_plot_cluster_n_clusters():
     geo = plot.plot(weights, n_clusters=3, show=False)
     assert isinstance(geo, DataGeometry)
 
-
-def test_plot_cluster_HDBSCAN():
-    # should return 10d data since ndims=10
-    geo = plot.plot(weights, cluster='HDBSCAN', show=False)
-    assert isinstance(geo, DataGeometry)
-
-
 def test_plot_nd():
     geo  = plot.plot(data, show=False)
     assert all([i.shape[1]==d.shape[1] for i, d in zip(geo.data, data)])