An R package which implements many measures for (graph) partition comparison. For more details, see the package details on CRAN (https://cran.r-project.org/package=partitionComparison)
TODO list
- Distance between one partition versus many partitions, vectorized computation (no for loop). Useful when there is 'true partition' and have posterior samples of partition.
- Bug in the classificaton error metric, when k = length(unique(p)) > k' = length(unique(q)).
- Binder loss and related loss functions missing
- Only accepting input as "Partition" S4 class makes package unaccessible.
- Canonical representation of partition. c(1,2,2,3) and c(2,3,3,1) should be the same when converted to the class "Partition".
0.2.5
- Changed maintainer information and updated URL/BugReports links
0.2.4
- Fixed some encoding issues
0.2.3
- Replaced
==comparisons by tolerance permissiveisTRUE(all.equal(x, y))
0.2.2
- First version published on CRAN
- Finalized description, documentation, references
- Added a
compareAllmethod to run comparison with all measures
0.2.1
- Dropped igraph dependency in favor of lpSolve
- Added some tests for method signature registration
- normalizedMutualInformation now with default "min"
- Rewritten registration of measure methods for plain partition vectors
0.2.0
- First finished version