Releases: moj-analytical-services/splink_graph
Releases · moj-analytical-services/splink_graph
v0.8.2 splink_graph update for spark 3.1.1 and graphframes 0.8.2
splink_graph updated so it works on contemporary version of Spark (3.1.1) and graphframes (0.8.2 )
Full Changelog: v0.5.0...v0.8.2
v0.5.0 documentation update and a few more metrics
What's Changed
- Create DOCUMENTATION.md by @EKenning in #81
- add num of articulation pts cluster metric by @mamonu in #79
- Create splink_graph_guide.md by @EKenning in #82
- Update splink_graph_guide.md by @EKenning in #84
- Mamonu docs patch 1 by @mamonu in #83
- Mamonu doc patch 2 by @mamonu in #85
- add graph_hashes with edge attributes by @mamonu in #86
New Contributors
Full Changelog: v0.4.21...v0.5
v0.4.21 latest version with fixes and some new cluster metrics
What's Changed
- Update TERMINOLOGY.md by @pratibha-vellanki in #66
- Assortativity cluster metric added by @mamonu in #67
- Fix problem with num_bridges issue 71 by @RobinL in #74
- remove rounding and simplify expressions by @RobinL in #75
- Degeneracy cluster metric added by @mamonu in #78
New Contributors
- @pratibha-vellanki made their first contribution in #66
Full Changelog: v0.4.19...v0.4.21
v0.4.19 latest version with small fixes
main additions from previous release:
- support for python 3.6 added and tested
- graph hash function has its own pandas_udf
version to be used in AWS Glue and ONS DAP
- small fixes
v0.4.12 added connected components
added connected components functionality
- graphframes (working in a distributed manner) version / needs jars (included) / util functions help setting up
- networkx (working on local master node) version since not every case needs scale
- nx based cc also saves optionaly each subcluster in networkx edgelist format (this is useful for embeddings ... tested and it works)
added number of bridges cluster metric
- number of bridges cluster metric is added. Useful for finding problematic clusters/subgraphs
v0.4.9 API for demos etc
Merge pull request #42 from moj-analytical-services/APIfixes fix some default parameter nonsense i had on edge_metrics
more consistent API
From this release owardn, functions to be seperated logically in appropriate submodules:
eg
from splink_graph.cluster_metrics import x
from splink_graph.node_metrics import x
from splink_graph.edge_metrics import x
Also column names are more consistent.
splink-graph 0.3.12
- added more cluster statistics functions
- added custom fields to pandas udf functions
- added more tests
Tidy up and consolidate cluster statistics functionality
- linting
- diameter_radius_transitivity function added in order to get most important cluster statistics
from one pandas_udf function