Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Dev/dendrogram #490

Open
wants to merge 2 commits into
base: master
Choose a base branch
from
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Jump to
Jump to file
Failed to load files.
Diff view
Diff view
55 changes: 55 additions & 0 deletions graphistry/compute/cluster.py
Original file line number Diff line number Diff line change
Expand Up @@ -436,3 +436,58 @@ def transform_dbscan(
)
return g
return emb, X, y, df



def get_dendrogram_edges(df: pd.DataFrame, as_graph: bool = True) -> Union[pd.DataFrame, Any]:
"""Converts a dataframe of feature embeddings to a dendrogram graph with edges between each merge
This will calculate what AgglomerativeClustering does under the hood, but using the linkage matrix

Args:
:df: dataframe of feature embeddings
:as_graph: whether to return a graphistry graph or a dataframe of edges
Usage:
::
g = graphistry.edges(edf, 'src', 'dst').nodes(ndf, 'node')
g2 = g.umap().dbscan() # or g2 = g.featurize()
g3 = get_dendrogram_edges(g2.get_matrix(), as_graph=True)
"""
from scipy.cluster.hierarchy import linkage
import graphistry

# df is the numeric dataframe from umap, or featurize
Z = linkage(df, 'ward')
# Convert to a DataFrame
df2 = pd.DataFrame(Z, columns=['src', 'dst', 'dist', 'size'])

# Create a new node for each merge
num_samples = len(df)
df2['src'] = df2['src'].astype(int)
df2['dst'] = df2['dst'].astype(int)

# The new node is the index + the number of samples
df2['new_node'] = df2.index + num_samples

# Convert the dataframe to have each edge as a row
edges_src = pd.DataFrame({
'node1': df2['new_node'],
'node2': df2['src'],
'dist': df2['dist'],
})

edges_dst = pd.DataFrame({
'node1': df2['new_node'],
'node2': df2['dst'],
'dist': df2['dist']
})

edges = pd.concat([edges_src, edges_dst])

# Handle data type
edges['node1'] = edges['node1'].astype(int)
edges['node2'] = edges['node2'].astype(int)

if as_graph:
g = graphistry.edges(edges, 'node1', 'node2')
return g
return edges
23 changes: 21 additions & 2 deletions graphistry/tests/test_compute_cluster.py
Original file line number Diff line number Diff line change
Expand Up @@ -4,7 +4,7 @@
import graphistry
from graphistry.constants import DBSCAN
from graphistry.util import ModelDict
from graphistry.compute.cluster import lazy_dbscan_import_has_dependency
from graphistry.compute.cluster import lazy_dbscan_import_has_dependency, get_dendrogram_edges

has_dbscan, _, has_gpu_dbscan, _ = lazy_dbscan_import_has_dependency()

Expand Down Expand Up @@ -67,7 +67,26 @@ def test_transform_dbscan(self):
g3 = g2.transform_dbscan(ndf, ndf, verbose=True)
self._condition(g3, kind)



class TestDendrogram(unittest.TestCase):

@pytest.mark.skipif(not has_dbscan, reason="requires ai dependencies")
def setUp(self) -> None:
g = graphistry.nodes(ndf).edges(edf, 'src', 'dst')
gs = []
for kind in ['nodes', 'edges']:
g2 = g.umap(kind=kind, n_topics=2, dbscan=False).dbscan(kind=kind, verbose=True)
gs.append(g2)
self.gs = gs

@pytest.mark.skipif(not has_dbscan, reason="requires ai dependencies")
def testDendrogramToGraph(self):
for kind, g2 in zip(['nodes', 'edges'], self.gs):
g3 = get_dendrogram_edges(g2.get_matrix(kind=kind))
self.assertTrue('node1' in g3._edges, 'dendrogram graph has no `node1` column')
self.assertTrue('node2' in g3._edges, 'dendrogram graph has no `node1` column')


if __name__ == '__main__':
unittest.main()