Evaluation of Dynamic Communities¶
This section contains functions useful to evaluate the quality of dynamic communities.
They were introduced in XXX.
- They can be split in 3 categories:
- Evaluation of an average value at each step (similarity_at_each_step,`quality_at_each_step`)
- Evaluation of smoothness (SM_L,`SM_N`,`SM_P`)
- Longitudinal evaluation (longitudinal_similarity)
A benchmark is also proposed that can be used to reproduce the results presented in the paper XXX.
Main evaluation functions¶
similarity_at_each_step(…[, score]) |
Compute similarity at each step |
quality_at_each_step(dynamicCommunities, …) |
Compute a community quality at each step |
SM_L(dyn_com[, sn_duration]) |
Smoothness for labels |
SM_N(dyn_com) |
Smoothness for nodes |
SM_P(dyn_com) |
Smoothness for partitions |
longitudinal_similarity(…[, score, …]) |
Longitudinal similarity |
Helper functions that could be used to evaluate smoothness¶
nb_node_change(dyn_com) |
Compute the total number of node changes |
entropy_by_node(dyn_com[, sn_duration, …]) |
Compute the entropy by node. |
consecutive_sn_similarity(dynamicCommunity) |
Similarity between partitions in consecutive snapshots. |
Benchmark¶
DCD_benchmark(methods_to_test, mus[, …]) |
Compute stats and running time for methods |